The World Trade Center site was worked around the clock from September 11, 2001 until a ceremony on May 30, 2002 closed the operation, which CNN described that day as eight months and nineteen days of work and 108,342 truckloads of debris. The people who worked there after the first morning were, with few exceptions, not in danger of being killed by the building, and many thousands of them have since been monitored or treated under a federal program for conditions associated with what they breathed. The recovery phase, rather than the day itself, is the part of this incident most likely to change how you run a long one.

Rescue to recovery, and why that changeover is hard to announce

Part two of this series dealt with the casualty surge that hospitals prepared for and that did not arrive, which is a planning failure of a completely different shape, and the reason it matters here is that the same absence of living casualties shaped what happened on the site for the next nine months. The classification question at the World Trade Center was not settled quickly, and the published record does not give a single clean moment at which the operation became a recovery. CNN, reporting on September 19, 2001, wrote that officials were continuing to refer to the work at the site as a rescue mission, emphasizing the possibility that people could still be found alive in the seven-level underground complex beneath the plaza, and that report is eight days after the last person was taken out of the debris alive.

What the record does show is a drawdown argued about in public rather than a declaration. The 9/11 Memorial and Museum’s published timeline records an announcement on October 30, 2001 that city officials planned to reduce the number of uniformed personnel working at the site so that staff could return to their assignments elsewhere in the city, and it records that personnel responded to that announcement, which is a polite way of saying the reduction was contested at the time. The same timeline records the departure of the last Federal Emergency Management Agency urban search and rescue task force from the site, and I could not pin that date to a source I was willing to publish, so I am leaving the date off rather than supplying one from a secondary summary.

The reason this transition is the hardest decision in the whole nine months is that it governs risk acceptance, and everybody on a collapse site understands that even if nobody says it out loud. An incident commander will accept a great deal of hazard to reach a living person, including sending people into voids, working under unshored steel and cutting in atmospheres that would stop the job on any ordinary day. Once the operation is a recovery, the same hazards buy a different thing, and the honest professional position is that the remains of the dead do not justify the exposures that a live rescue justifies, which is a sentence that is easy to write in an office and very difficult to say at a fence line where families are standing.

The announcement is also heard as a verdict on the people still missing, which is why agencies delay it, and the delay has a cost that shows up later in the medical record rather than immediately in the casualty count. If the classification does not change, the risk posture does not change, and a site that is still nominally in rescue is a site where slowing down to fit a respirator or to wait for a monitoring reading feels like a betrayal of someone who might be under the next slab. That mechanism is the part worth carrying away from this section, because it operates identically at a two-story residential collapse in your county, where the same reluctance to say the word recovery will keep crews working in a torch-and-cut atmosphere for hours after the technical basis for doing so has gone.

The people taken out alive, and how the count gets garbled

The number in general circulation is twenty. An account by two retired New York City police officers, published by Northwell Health, states that only twenty people survived the collapse of the twin towers and were pulled from the rubble, and the two officers describe their own part in the rescue of Port Authority police Sergeant John McLoughlin and Officer Will Jimeno, who were trapped under roughly thirty feet of debris. I was not able to verify that figure against an official roster maintained by a city or federal agency, and until somebody produces one, the defensible way to state it is that twenty is the figure consistently published in accounts by participants and by the institutions that have collected their accounts.

The last live rescue is better documented than the total. Genelle Guzman-McMillan, an administrative assistant employed by the Port Authority of New York and New Jersey, was located by a search dog and taken out of the debris on September 12, 2001 after roughly twenty-seven hours trapped, and her own memoir, Angel in the Rubble, is the primary account. The date is routinely garbled in circulation, and I found outlets whose headlines place the rescue on September 11 while their own text describes twenty-seven hours under the debris, which is arithmetically impossible for a collapse that occurred mid-morning on the eleventh. If you teach this event, put September 12 on the slide and be ready to explain the twenty-seven hours, because someone in the room will have read the wrong version.

Setting those two facts beside each other produces the finding that matters operationally. Every documented live rescue from that debris field happened inside roughly the first day, and the operation continued to be described publicly as a rescue for weeks afterwards, which means the gap between the end of live rescue and the end of rescue classification was measured in weeks rather than in hours. I am not offering that as a criticism of any individual, because the decision to announce the end of hope is not a technical decision and the people making it were under pressures no technical manual addresses. It is a planning fact, and the planning response to it is to decide in advance who holds the authority to reclassify, on what advice, and what changes about site operations the moment it happens.

The site as a workplace: duration, tonnage and who was on it

The physical scale is documented well enough to state with sources attached, and the sources do not entirely agree. Construction Equipment Guide, reporting on the closing ceremony, put the removal at 1.7 million tons of wreckage, while the 9/11 Memorial and Museum’s page on the May 30, 2002 commemoration gives about 1.8 million tons over nine months and its page on the rescue and recovery community describes an operation that removed two million tons. Those three figures come from two organizations and they disagree, probably because the boundary between structural debris, soil and material removed during later excavation was drawn differently in each case, so the honest range to publish is roughly 1.7 to 2 million tons with the publisher named. CNN’s report of the closing ceremony gives 108,342 truckloads and notes that the job had originally been estimated to take about a year, with former Mayor Rudy Giuliani quoted that day saying the work was done “much quicker and more effectively than anyone had a right to expect”.

The site burned for a long time, which changed the atmosphere every worker was standing in. The 9/11 Memorial and Museum’s description of the rescue and recovery community refers to fires that burned for over one hundred days alongside the debris removal, and a fire of that duration in a mixed fuel load of building contents, fuel and electrical materials produces combustion products continuously rather than as a single release event. Work went on around the clock, and Construction Equipment Guide describes the project in exactly those terms, which means the exposure question is not about one bad shift but about repeated shifts over months, with the same individuals returning to the same ground after a rest period too short to clear anything from their airways.

The workforce was mixed in a way that almost no agency plans for. Fire, police and Port Authority police personnel worked alongside federal urban search and rescue task forces early on, and as the operation became a heavy demolition and debris removal job it drew in ironworkers, operating engineers, laborers, truck drivers, sanitation and utility crews, medical examiner staff, chaplains, medical volunteers and people who simply arrived and were useful. Material was moved to the Fresh Kills landfill on Staten Island for sorting, and the identification effort that ran from that sorting continued for years and is covered separately on this site in the piece on mass fatality management, which is where the medical examiner’s authority and the family assistance side of this belong.

The headcount is where I will not give you a number. No contemporaneous roster of everyone who worked on that site exists as far as I could establish, published estimates differ because they define the population differently, and the enrollment figures of later health programs count people who came forward rather than people who were there. The one enrollment figure I can attribute is from a public safety timeline published by Police1, which states that the World Trade Center Health Registry received federal funding in July 2002 and that more than 71,000 people enrolled during its initial survey period, including rescue and recovery workers, residents and others caught in the dust cloud. That is an enrollment count across several exposed populations, so anyone who converts it into a site headcount is inventing a statistic, and the mechanism that made a real headcount impossible is the same one your agency will face on day four of anything large, which is that people arrive, work and leave without anyone recording that they were present.

The dust, what was said about it, and which body said it

The composition of the dust is described in the primary document most people reach for last. The Environmental Protection Agency’s Office of Inspector General, in its report titled EPA’s Response to the World Trade Center Collapse: Challenges, Successes, and Areas for Improvement, Report No. 2003-P-00012, dated August 21, 2003, describes airborne dust from the collapse blanketing Lower Manhattan and dispersing into surrounding offices, schools and residences, and characterizes it as a complex mixture of building debris and combustion by-products containing such ingredients as asbestos, lead, glass fibers and concrete dust. That is a federal oversight body describing the hazard in its own words two years afterwards, and it is a better citation than any secondary account of what was in the air.

What was said publicly is documented and it is contested, so it needs to be stated precisely and attributed rather than summarized. EPA issued reassurances about environmental conditions in Lower Manhattan beginning in mid-September 2001, including a press release dated September 18, 2001 in which the then Administrator, Christine Todd Whitman, stated that monitoring results indicated the air near the World Trade Center was safe to breathe, and that press release and its date are the anchor for everything argued since. The OIG’s 2003 report concluded that when the agency made that September 18 announcement it “did not have sufficient data and analyses to make such a blanket” statement, and the report recommended that the agency develop procedures for emergency risk communication so that public pronouncements about health risk are supported by available data and analysis. The same report states that New York City in fact exercised the lead role on indoor air, that EPA began to assume a lead role in February 2002 after the city was criticized, and that the full extent of public exposure to indoor contaminants is unknown.

A second OIG product in 2003, a survey of air quality information related to the World Trade Center collapse, examined how the agency communicated through press releases, interviews and its website, and framed its own purpose around concerns that had been raised about government communications and “the impact these communications had on the actions taken by the public and responders” to reduce exposure. The OIG report also dealt with the involvement of the White House Council on Environmental Quality in the review of draft press releases, and I am pointing you to the report itself rather than summarizing that part, because it is the most contested section of the record and it deserves to be read rather than paraphrased by me. This is not a place to assign a motive to any individual, and the point that survives without any motive attached is that a public statement about ambient air in a district is not a statement about the working atmosphere on a burning debris pile, and workers on that pile heard the two as the same message.

The litigation and the later public record should be handled with the same care. WNYC reported that a federal judge allowed a class action brought in 2004 by residents, students and area workers to proceed against the agency and the former Administrator, and characterized the court’s description of the statements in strong terms, and I did not verify the final disposition of that case on appeal for this piece, so check the case history before you repeat any judicial language. In September 2026 the City of New York released a very large volume of records relating to air quality after the attacks, reported by Fox News at more than 170,000 pages, and that material is new enough that no independent technical review of it had been published at the time I wrote this, which is exactly the point at which a careful writer says so and waits.

Respiratory protection and the reasons it was worn inconsistently

Respirators were available on that site, distributed in very large numbers, and the photographic record of the nine months shows people working with half masks, with filtering facepieces, with surgical masks that do nothing useful against fibers, with bandanas and with nothing at all, frequently within the same frame. I did not verify a distribution total for this piece and I am not going to publish one, because the number that circulates is usually a count of units handed out rather than a measure of protection achieved, and those two quantities are not related to each other in any simple way. The useful question for a working agency is not how many respirators arrived at the site but what fraction of the person-hours worked there were worked inside a properly selected, properly fitted and properly maintained respirator, and nobody can answer that for the World Trade Center because nobody was measuring it.

The reasons protection was worn inconsistently are documented in general terms across the published accounts and they are all familiar to anyone who has supervised long-duration work. A half-face air purifying respirator requires a fit test against the specific make and size, and it will not seal against facial hair, so the workforce that arrives from trades and agencies with no respiratory program is unfittable at the moment it arrives. Cartridges have a service life that depends on the contaminant and the work rate, so sustaining protection across a nine-month operation is a logistics problem of change-out schedules and resupply rather than a one-time issue of handing out masks. Heat and exertion make a tight-fitting facepiece punishing during heavy manual work, and speech through a half mask is difficult enough that people on a noisy debris pile pull it down to talk to the person beside them or to use a radio, which is the single most common breach I have watched in twenty-five years of fireground and site work.

Above all of that sits the message environment described in the previous section, because the OIG’s own survey was framed around the effect of government communications on the actions responders took to reduce their exposure, which is an oversight body saying in its opening pages that what was said in public plausibly affected what people put on their faces. The culture layer is the part no report captures cleanly, and it is the part I will state as my own professional judgment rather than as a finding, which is that on a site where people believe the work is a debt owed to the dead, protective equipment gets read as a statement about your commitment and the person who keeps the mask on becomes the person who is asked why.

The structural problem underneath all of it is that respiratory protection is enforced by an employer, and a site with hundreds of employers and a large number of people with no employer of record has no single party who can be held to a respiratory protection program. A fire department can order its own members into a respirator and discipline them for not wearing it, because there is a chain of command and an occupational health program behind that order. A volunteer who drove in from three states away, a contractor’s day laborer and an off-duty member from another jurisdiction working outside his own department’s control sit outside every one of those mechanisms, and unless the site has a single safety authority with the power to deny access, the protection program that exists on paper governs only a fraction of the people breathing the air.

The health record and the federal program, handled carefully

Medical follow-up began during the operation rather than after it. The 9/11 Memorial and Museum’s timeline records that the New York City Fire Department established a medical monitoring program for participating firefighters and EMS members in order to track their health, and the World Trade Center Health Registry, according to the Police1 public safety timeline, received federal funding in July 2002 and enrolled more than 71,000 people in its initial survey period across responders, residents and others who were in the dust cloud. A registry built on voluntary enrollment is a different instrument from a roster built at the gate, and the epidemiology that followed has had to work within that limitation, which the registry’s own published documentation describes better than I can.

The federal program that now provides monitoring and treatment has a name and a statutory history worth getting right, because it is misnamed constantly. The James Zadroga 9/11 Health and Compensation Act of 2010, Public Law 111-347, was signed on January 2, 2011, and Title I of that act amended the Public Health Service Act to establish the World Trade Center Health Program within the Department of Health and Human Services, providing medical monitoring and treatment to eligible responders, meaning firefighters and related personnel, law enforcement officers and rescue, recovery and cleanup workers at the New York City site, at the Pentagon and in Shanksville, Pennsylvania, and also establishing benefits for eligible survivors, meaning people present in the dust or dust cloud on September 11, 2001 or who worked, resided or attended school, childcare or adult daycare in the New York City disaster area. Those definitions are from the program’s own published Laws page, and they matter because eligibility turns on them.

The reauthorization history is equally specific. On December 18, 2015, the Consolidated Appropriations Act, 2016, Public Law 114-113, incorporated the James Zadroga 9/11 Health and Compensation Reauthorization Act and reauthorized the program for seventy-five years, through 2090, replacing the original five-year funding period that would have ended in 2016. On July 29, 2019, the Never Forget the Heroes: James Zadroga, Ray Pfeifer, and Luis Alvarez Permanent Authorization of the September 11th Victim Compensation Fund Act, Public Law 116-34, dealt with the separate Victim Compensation Fund, and on September 27, 2019 the Continuing Appropriations Act, 2020, and Health Extenders Act of 2019, Public Law 116-59, raised the numerical limits on enrollment of responders and survivors in the health program. The program’s Laws page also records an amendment on February 3, 2026, in the Consolidated Appropriations Act, 2026, Public Law 119-75, changing the formula used to calculate the program’s funding, and since that is recent I would check the current page rather than rely on this paragraph a year from now.

On the health findings, I am going to give you citations rather than numbers, because the numbers belong in the papers and I could not re-verify specific effect sizes to the standard this article requires. The New York City Fire Department cohort produced the earliest peer reviewed work, including Prezant and colleagues in the New England Journal of Medicine in 2002 on cough and bronchial responsiveness among firefighters who worked at the site, and Aldrich and colleagues in the same journal in 2010 reporting on lung function in rescue workers several years after the exposure, and Zeig-Owens and colleagues in The Lancet in 2011 on cancer incidence in that cohort. The strength of the evidence is not uniform across conditions, and it is worth saying plainly which is which, since for upper and lower airway disease and for gastroesophageal reflux the exposure relationship is supported by repeated cohort findings with a gradient by arrival time and intensity of exposure, while for individual cancer sites much of the published work reports associations in cohorts with real limitations around comparison populations and surveillance intensity, which the authors themselves discuss. Certification of a condition by the health program is an administrative determination made under published criteria and a published list of covered conditions, and it is not a finding that a particular exposure caused a particular person’s illness, which is a distinction the program’s own documentation handles carefully and a great many news accounts do not.

The doctrine that followed, and whether a mid-sized agency does any of it

The doctrinal response to all of this is now unremarkable on paper. A long-duration site is treated as an occupational health problem with an industrial hygiene component, meaning a site safety and health plan, a designated safety officer with assistants covering the operational periods, air monitoring appropriate to the contaminants present, a respiratory protection program with medical evaluation and fit testing behind it, decontamination at the exit point, and a record of who was exposed to what. The federal respiratory protection standard administered by the Occupational Safety and Health Administration requires the written program, the medical evaluation, the fit test, the training and the cartridge change-out logic, and I am deliberately not citing the section number here because that is exactly the kind of detail people get wrong from memory. The National Fire Protection Association publishes consensus standards on fire department occupational safety and health programs and on medical requirements for members, and your training officer can produce the current editions faster than a search engine can.

The honest question is whether any of that actually happens at a mid-sized agency on a two-week incident, and my answer, from twenty-five years around this work, is mostly no. The Safety Officer position gets filled on day one because the position is on the organization chart and somebody senior is standing there, and by day ten the person who filled it has gone back to their regular job and the position exists only as a name in an old incident action plan. Exposure records, if they exist at all, consist of unit logs that record which crews were assigned where, in a format that nobody has ever tried to search by individual, held on a server that will be migrated twice before anybody asks for them.

Fit testing is the clearest test of whether an agency is serious, and it is usually the clearest evidence that it is not. Most fire departments fit test their own members annually because their self-contained breathing apparatus program requires it, and most of those same counties have never fit tested the public works crews who will run the loaders at a debris site, the sheriff’s deputies who will staff the perimeter, the per diem EMS staff, the building inspectors who will do the damage assessment or the animal control officers who will work the same ground. When those people are committed to a debris operation, they are committed without protection that fits, which means the agency’s respiratory protection program covers the population least likely to be there in week three.

There is one piece of institutional history I am not going to retell here, because the Communications section of this site is running its own sequence on the policy record after 2001, including the national incident management structure and the spectrum argument, and duplicating that would add nothing. What belongs in this article is the narrower point that the safety officer function and the exposure record are the two parts of that structure that fail quietly, because nothing breaks when they are absent and the consequence shows up in somebody’s pulmonary function test fifteen years later, by which time the incident is closed, the agency has a different chief and the record that would have supported a claim was never created.

Why people take the mask off, and when that gets solved

Nobody on that site removed a respirator because they failed to understand that pulverized building material is bad for you. They removed it because it did not seal, because the cartridge was spent and the resupply was two blocks away, because they could not make themselves understood on a radio or to the person working beside them, because the heat inside the facepiece during heavy manual labor was genuinely intolerable, and because the social cost of being the one person still wearing it was higher than the perceived risk of taking it off. Every item on that list is a solvable engineering, logistics or supervision problem, and not one of them is solvable during the incident, which is the whole argument of this article compressed into a paragraph.

Fit is solved by testing everyone who could plausibly be committed, by stocking more than one make and size, and by having a written rule on facial hair that the agency is actually willing to enforce, since a rule that is waived under pressure is a rule that has already failed. Communication is solved by buying respirators compatible with the radio interface you already own, or by buying the accessory that makes them compatible, and by testing that combination during training rather than discovering at the site that the push-to-talk cannot be operated with a gloved hand while the mask is in place. Heat is solved by work and rest cycles written into the site safety plan with a named person responsible for calling them, and by rehabilitation that is set up before the first crew needs it. Supply is solved by knowing your daily cartridge burn rate at full commitment and by having a resupply arrangement that does not depend on a single vendor answering the phone on a Sunday.

Culture is the one that takes longest and it is the only one that cannot be purchased. It changes when officers wear the equipment without being told to, when the safety officer is a respected operational person rather than the member nobody else wanted to assign, and when the agency treats exposure documentation as part of finishing the job in the same way it treats putting equipment back in service. A chief who steps onto the pile without a respirator has set the standard for that site regardless of what the safety plan says, and that is worth stating because it is the cheapest intervention available to any agency reading this.

The people at greatest cumulative risk at a long incident are rarely the first-due companies, whose exposure is intense but short and who work inside an agency with a respiratory program. They are the people on the site in week three, who came from public works, from a contractor, from a neighboring county on a mutual aid request or from no organization at all, whose employer has no respiratory program, whose names were never written down against a location, and who will be trying to prove twenty years later that they were there. Building the record that those people will need costs almost nothing during the incident and cannot be reconstructed afterwards at any price, which is the practical inheritance of the nine months on the pile.

What to do at your agency

• Have your training officer produce the current fit test roster and compare it against a list of every group that could be committed to a debris site, naming public works equipment operators, jail and court security staff, per diem EMS personnel, building inspectors and animal control, and report the gap as a number at the next staff meeting.
• Ask your logistics officer to calculate the daily cartridge and filter consumption for a full-commitment operation lasting fourteen days, then confirm with your current vendor in writing how much of that they can deliver in forty-eight hours on a weekend.
• Write one paragraph into the safety annex of the emergency operations plan you already have that names who fills the Safety Officer position for operational periods after the first seventy-two hours, names at least two deputies by position title, and states that the position is staffed for the duration rather than for the first day.
• Pick a recent incident that ran longer than three shifts and ask your records custodian to produce, in a week, a list of every person who worked on it, where each one worked and on which days, and treat the result as the true measure of whether your agency could document an exposure twenty years from now.
• Telephone your occupational medicine provider and ask what a baseline respiratory surveillance record would involve per member, what their record retention period is, and what they would need from the incident to link an exposure to an individual, then file the answers with the date on them.
• Add one item to the agenda of a meeting that already exists, such as your county safety committee or your fire chiefs association, requiring a decision on who has authority to deny site access to unaffiliated volunteers and self-deployed responders who cannot be fit tested.
• Have your fire chief or emergency manager state in writing the trigger and the authority for reclassifying an incident from rescue to recovery, including who is consulted and what changes about risk acceptance at that moment, and put it in the same plan as the search and rescue annex.

Takeaways

• The last live rescue at the World Trade Center was Genelle Guzman-McMillan on September 12, 2001 after about twenty-seven hours in the debris, and the figure of twenty people taken out alive is consistently published in participant accounts, including the one carried by Northwell Health, but I could not verify it against an official agency roster.
• CNN reported on September 19, 2001 that officials were still describing the work as a rescue mission, which places the end of rescue classification weeks after the last live rescue, and the 9/11 Memorial and Museum’s timeline records an October 30, 2001 announcement to reduce uniformed staffing at the site that was publicly contested.
• The operation closed with a ceremony on May 30, 2002, described by CNN that day as eight months and nineteen days of work and 108,342 truckloads, and published tonnage figures disagree, running from 1.7 million tons per Construction Equipment Guide to about 1.8 million and two million tons on different pages published by the 9/11 Memorial and Museum.
• No contemporaneous roster of the site workforce exists that I could find, and the World Trade Center Health Registry enrollment of more than 71,000 in its initial survey period, as reported in the Police1 public safety timeline, counts several exposed populations who came forward rather than people who were present.
• The EPA press release of September 18, 2001 stated that monitoring indicated the air near the site was safe to breathe, and the agency’s Office of Inspector General concluded in Report No. 2003-P-00012, dated August 21, 2003, that the agency “did not have sufficient data and analyses to make such a blanket” statement.
• Respiratory protection failed on that site for reasons of fit, supply, heat, speech intelligibility and social cost rather than for lack of understanding, and every one of those is fixed before an incident or not at all.
• The World Trade Center Health Program was established by Title I of the James Zadroga 9/11 Health and Compensation Act of 2010, Public Law 111-347, signed January 2, 2011, and was reauthorized through 2090 by the Consolidated Appropriations Act, 2016, Public Law 114-113, signed December 18, 2015, with later amendments affecting enrollment limits and funding that are listed on the program’s own Laws page.
• Certification of a condition under that program is an administrative determination against published criteria rather than a finding that a specific exposure caused a specific person’s illness, and the strength of the underlying evidence is greater for airway and digestive conditions than for individual cancer sites.