At 5:04 p.m. on 17 October 1989 the Loma Prieta earthquake struck the San Francisco Bay Area. The deadliest single failure was about 60 miles from the epicenter, in West Oakland, where the upper deck of the double-deck Cypress Street Viaduct on Interstate 880 fell onto the lower deck. Of the 63 deaths most commonly attributed to the earthquake, 42 occurred on that structure. This guide covers why it failed on soft ground, the rescue that neighborhood residents started before organized crews arrived, the World Series myth, and the urban search and rescue and retrofit work that followed.

17 October 1989: a rupture in the Santa Cruz Mountains and damage across the region

The earthquake began beneath the Santa Cruz Mountains, near the peak called Loma Prieta, on a section of the San Andreas fault system southeast of the Bay Area’s population centers. Strong shaking lasted roughly 15 seconds in the epicentral area. The US Geological Survey gives the magnitude as 6.9 on the moment magnitude scale. Many news reports at the time gave 7.1, which was a surface-wave magnitude. The two numbers describe the same earthquake measured in different ways, and anyone comparing older and newer sources will see both. I use 6.9 because it is the figure the USGS publishes today.

Damage was spread unevenly across a large area. Downtown Santa Cruz and Watsonville, close to the rupture, lost many older unreinforced masonry buildings. In San Francisco the Marina District, much of it built on fill placed over former bay and lagoon, suffered liquefaction, building collapses and a major fire. A section of the upper deck of the eastern span of the San Francisco-Oakland Bay Bridge dropped onto the lower deck, and one motorist died there. The bridge was closed for about a month. Elevated freeways in San Francisco were damaged badly enough that several were later torn down rather than repaired.

The USGS gives the death toll as 63 and the number of injured as 3,757, with property damage of roughly 6 billion dollars. A few sources give a death toll slightly above or below 63, and other published damage estimates differ from the USGS figure, which is normal for a disaster of this size. I use the USGS numbers throughout and name them as such. The figure for the Cypress structure most commonly given in the histories and in Caltrans and USGS material is 42 deaths.

That concentration is the point of this guide. Two-thirds of the people killed in a regional earthquake died on a single structure, built to the design standards of its time, standing on the worst kind of ground in the area and roughly 60 miles from the epicenter. The pattern is the same one the site’s article on Mexico City 1985 describes for lake-bed clay, here applied to one transportation structure rather than a city district. Distance from the fault was a poor guide to where the deaths would be, because the soil under the structure and the way it was built mattered more.

The Cypress Street Viaduct: a 1950s double-deck freeway through West Oakland

The Cypress Street Viaduct carried the Nimitz Freeway, Interstate 880, through West Oakland toward the approaches of the Bay Bridge. It was a reinforced concrete structure with two decks, one direction of traffic on each, supported by a long series of two-level concrete frames called bents. It was designed in the early 1950s and opened in 1957. Each bent carried the lower deck on its lower beam and the upper deck on a frame that sat above it, so the strength of the joint between the lower and upper portions of each bent decided whether the upper deck stayed up.

The route ran through a historically Black, working-class neighborhood, and the elevated structure divided it physically for more than thirty years. That history matters to this story twice. The people who lived beside the structure were the first to reach it after it fell, and their objections later shaped the decision not to rebuild the freeway on the same alignment, which I cover in the final section.

The structure predated the design changes that California made after the 1971 San Fernando earthquake, which damaged freeway structures in the Los Angeles area. Caltrans revised its bridge design standards after 1971 and began a retrofit program, and the first phase of that program concentrated on restraining bridge spans at their expansion joints so they would not slide off their supports. Strengthening the columns of existing structures was a later and much larger undertaking, and in October 1989 the Cypress structure’s columns and joints had not been strengthened.

When the shaking reached Oakland, the upper deck along a long stretch of the viaduct came down onto the lower deck, which accounts commonly describe as more than a mile of the structure. In many places the upper deck settled directly on the lower deck with vehicles crushed between them. In others it came to rest at an angle or on debris, leaving gaps of a few feet in which people in their cars survived. A portion of the structure at one end did not collapse, and that difference turned out to be the clearest single clue to why the rest of it failed.

Bay mud and column joints: why the upper deck came down

Post-earthquake investigations identified two factors working together, and neither alone explains the collapse. The first was the ground. The portion of the viaduct that collapsed stood on soft bay mud and artificial fill near the shoreline, while the portion that stood was on firmer soil. Soft, saturated sediments amplify certain frequencies of earthquake shaking, so a structure on bay mud can experience much stronger motion than one a short distance away on firmer ground. Strong-motion recordings made across the Bay Area during this earthquake showed that amplification at soft sites, and it is the same effect that contributed to the damage in the Marina District.

The second factor was the detailing of the concrete. The upper-deck columns sat on short sections, often called pedestals, above the lower deck. As the investigations described it, the reinforcing steel through those joints lacked the continuity and confinement that later codes require, so when the frames were pushed back and forth by repeated cycles of strong motion the concrete at the joints cracked and broke apart in shear rather than bending and holding. Once those joints failed, the upper columns lost their support and the upper deck dropped. Engineers call this kind of construction non-ductile, meaning it has little capacity to deform without breaking, and it was common in concrete structures designed before the 1970s.

Governor George Deukmejian appointed a Board of Inquiry chaired by George W. Housner, a Caltech engineer, to investigate the transportation structure failures. The board’s 1990 report, titled Competing Against Time, examined the Cypress collapse in detail and called for the state to treat the seismic safety of its transportation structures as a high and urgent priority. Readers who need the engineering findings themselves should go to that report and to the technical papers written by the investigators, which describe the joint failures far more precisely than I can here.

The practical lesson for anyone outside structural engineering is that the risk was a combination of two conditions that were each knowable in advance. The soils along the shoreline had been mapped, and the general weakness of non-ductile concrete was understood in the engineering community by the 1980s. What was missing was money and time to strengthen every vulnerable structure before the next large earthquake, which is the problem the Housner report’s title refers to.

The World Series myth: what the game actually changed on the roads

The earthquake struck about half an hour before Game 3 of the 1989 World Series between the Oakland Athletics and the San Francisco Giants was due to start at Candlestick Park in San Francisco. The pregame broadcast was live on national television when the shaking began, and the signal went off the air. Candlestick held up, the crowd was evacuated, and the series resumed ten days later. Because the event reached a national audience in real time, it was associated with baseball from the first minute, and several retellings have carried that association further than the record supports.

The myth, and it is a myth, is that the World Series crowd was caught on the Cypress structure and that game traffic accounts for the deaths there. The stadium was in San Francisco, the game was about to begin, and the people going to it were largely at or near Candlestick by 5:04 p.m. I have not seen any reliable account that identifies the Cypress dead as a ballgame crowd. The people on the viaduct were, as far as the record shows, ordinary weekday traffic on a major commuter route.

The real effect of the game, as it is usually described, ran in the other direction. Contemporary news coverage and officials widely reported that many people in the region had left work early or stayed home to watch the game, so the evening commute was lighter than on a normal Tuesday. That account is the standard explanation for why the death toll on the Cypress structure was lower than first feared, and it is plausible on its face. I have not seen a count of vehicles on the structure at the moment of collapse compared with a typical weekday, so I cannot say how much lighter traffic was or how many lives the difference saved, and I would treat any precise figure for that as unsupported.

The other consequence of rush hour expectations was the early casualty estimate. In the first day or two, some reports put the toll on the Cypress structure and in the earthquake overall far higher than the final count, in the hundreds in some broadcasts. Those estimates were built partly on assumptions about normal rush hour volume and were revised downward over several days as crews worked through the collapsed sections and recovered the dead. The final Cypress figure most commonly cited is 42.

The first hours: neighbors and nearby workers on an unstable deck

The people who reached the collapsed viaduct first were the people who lived and worked beside it. Accounts from the time and the later histories describe West Oakland residents running toward the structure within minutes, climbing onto the lower deck and the wreckage with ladders, and pulling injured motorists out of cars. Workers from nearby industrial businesses are described bringing ladders, forklifts and other equipment to reach the lower deck and the gaps between the decks. The general account of neighborhood residents and nearby workers reaching trapped people before organized rescue was established along most of the collapse is consistent across the sources I have read, although the details of individual rescues vary between retellings and I am not going to repeat any of them as settled.

The conditions those volunteers worked in were dangerous by any professional standard. The upper deck had already failed once and could shift further, aftershocks followed the main shock, and the crushed vehicles held fuel. Light was failing, because the earthquake struck in the early evening in October. None of the people climbing onto the structure in the first hour had any way to know which sections were stable, and nobody was keeping track of who had gone into the wreckage or come back out of it. I have not seen a reliable account of volunteers killed or seriously injured on the structure, and I would not assume either way without one.

Several accounts describe friction when official agencies took control of the scene and moved volunteers back. That is the ordinary and probably necessary course at a collapse once trained rescuers, engineers and a command structure are present, but it was felt by residents who had been doing the work and who then watched from behind a line. Later writing about the neighborhood treats the residents’ response as a source of pride and the transition as a grievance. Both readings appear in the record, and I present them as the accounts describe them rather than judging who was right in particular cases.

For a working agency, the useful fact is structural rather than moral. At a sudden mass casualty collapse in a populated area, the first people on scene are neighbors and passersby, and they will already be inside the hazard area when the first engine arrives. The same pattern appears in the site’s articles on Mexico City 1985 and Kobe 1995. A plan that assumes the scene will be empty until responders secure it describes a situation that has not happened at any major urban collapse I know of.

The organized rescue: working between the decks and the late live find

Oakland Fire Department crews, police, the California Highway Patrol and Caltrans personnel arrived in the first hours, followed by mutual aid companies and heavy equipment. The work went on through the night under portable lighting. The physical problem was unusual even for rescuers experienced in building collapse. Victims were in vehicles sandwiched between two massive concrete decks, sometimes in a gap of only a few feet, along a linear site more than a mile long, so rescuers had to crawl into the space between the decks and work around crushed metal with hand tools and hydraulic equipment while engineers judged whether the deck above them would hold.

Caltrans engineers and other structural specialists assessed the bents, and crews placed shoring and cribbing to stabilize sections before entry. Physicians from area hospitals came to the scene, and accounts describe surgeons working in the confined space between the decks where patients could not be freed in any other way. As the hours passed, the number of people found alive fell sharply, and the work shifted from rescue to recovery, with cranes removing sections of the upper deck so that vehicles and bodies could be reached.

The rescue most remembered from the Cypress structure came after that shift. On Saturday, 21 October, four days after the earthquake, workers found a man named Buck Helm alive in his car between the decks, and his rescue after roughly 90 hours trapped was widely reported. He died in the hospital about a month later, according to the reporting at the time. His survival is often cited in collapse rescue training as evidence that a void space can sustain life for days, and that the transition from rescue to recovery should be treated as a deliberate command decision with a stated basis rather than something that happens because the crews are exhausted.

The organized response at Cypress also showed the limits of what a fire department of the time carried for heavy concrete collapse. Hydraulic rescue tools designed for vehicle extrication, ordinary ladders and portable lighting were the equipment on hand in the first hours, and moving decks weighing many tons required cranes and construction contractors. That gap, between local fire department capability and what a large concrete collapse demands, is the one that dedicated urban search and rescue teams exist to fill.

Urban search and rescue after 1989: what Loma Prieta did and did not start

Loma Prieta is often described as the event that created urban search and rescue in the United States, and that overstates it. The need had been shown by the Mexico City earthquake of 1985 and the Armenian earthquake of 1988, to both of which American fire department teams from Fairfax County, Virginia and Metro-Dade County, Florida deployed. FEMA dates its National Urban Search and Rescue Response System to 1989, the year of Loma Prieta, and the system grew through the early 1990s. I have not seen a primary document that makes Loma Prieta the single cause of the federal system, and it is more accurate to treat it as one of several disasters that pushed the same way.

In California the effect was more direct. The state Office of Emergency Services, now the California Governor’s Office of Emergency Services, built up a state urban search and rescue program in the following years, and California fire departments sponsor a large share of the federal task forces. Of FEMA’s 28 national task forces, eight are based in California, and the Oakland Fire Department is the sponsoring agency for one of them. The Cypress collapse is commonly cited in California fire service histories as a driver of that investment, and the 1994 Northridge earthquake and the 1995 Oklahoma City bombing tested the system that resulted. The site’s article on Oklahoma City covers the federal task force deployment there.

One common confusion should be cleared up while the subject is open. California’s Standardized Emergency Management System, the statewide framework built on the Incident Command System, came out of the 1991 Oakland Hills fire, not Loma Prieta. The two disasters struck the same city two years apart, and retellings sometimes merge their consequences, but the legislation establishing SEMS followed the fire.

The doctrinal change that matters most for a local agency is the tiered approach to collapse rescue. National consensus standards for technical rescue and FEMA training now distinguish awareness, operations and technician levels of capability, so that the first-arriving engine company knows what it can safely do, such as surface rescue of lightly trapped victims, scene control and size-up, and what it must hold for a specialized team. At Cypress in 1989, that line was drawn on the spot by whoever was in charge of each section of the structure.

The Housner board, column jackets and a freeway that was not rebuilt

The Housner board’s report led to a 1990 executive order from Governor Deukmejian declaring it state policy that transportation structures be seismically safe, and Caltrans accelerated its retrofit program. The most visible technique was steel jacketing of concrete bridge columns, in which a steel shell is placed around an existing column and the space between them grouted, so that the column is confined and can deform without the concrete breaking apart. Retrofit work extended to joints, footings and the connections between spans. In 1996 California voters approved a statewide bond measure for seismic retrofit of bridges, and readers who need its terms should look it up through the California Secretary of State’s ballot measure records.

The state’s toll bridges were handled as a separate program. After study, the eastern span of the Bay Bridge, the one whose upper deck section had fallen in 1989, was replaced rather than retrofitted, and the new span opened in 2013. In San Francisco, the damaged Embarcadero Freeway was demolished in 1991, and other elevated freeway structures in the city were later removed or replaced, which reshaped parts of the waterfront and the neighborhoods around them.

The Cypress structure itself was not rebuilt where it stood. West Oakland residents, who had lived beside the viaduct since the 1950s, pressed for the replacement freeway to be routed away from the neighborhood, and Caltrans built the replacement on a new alignment around the residential area, opening in 1997. The old route became Mandela Parkway, a landscaped boulevard, and a memorial park near the former structure commemorates the people who died there and the residents who went onto the wreckage to help.

The 1994 Northridge earthquake provided an early test of the retrofit work. Several freeway structures in the Los Angeles area collapsed in that earthquake, and the post-earthquake reviews generally reported that retrofitted bridges performed well while the collapses were concentrated in structures that had not yet been strengthened. Anyone citing the performance of specific structures should go to the Caltrans and FEMA post-Northridge reports rather than to summaries, because the details vary by structure and the general statement is only as good as the record behind it.

What to do at your agency

• The emergency management coordinator asks the state department of transportation district office for the list of bridges and elevated roadways in the jurisdiction with their seismic retrofit status, compares it against the state geological survey’s soft-soil or liquefaction map, and records in the hazard mitigation plan which structures meet both conditions.
• The fire department’s special operations or training officer confirms in writing which collapse rescue resource, whether a regional technical rescue team, a state team or a FEMA task force, the mutual aid plan calls for, how it is requested, who can request it and its estimated arrival time, and adds that paragraph to the technical rescue annex.
• The volunteer coordinator writes one paragraph into the existing volunteer management annex covering spontaneous rescuers at a collapse scene, naming a check-in point, a perimeter debrief to capture what they have seen and a safety briefing before anyone who stays is given a task.
• The communications center manager checks whether the computer-aided dispatch system has a run card for a structural or elevated roadway collapse, and if it does not, writes one that dispatches heavy rescue, a structural specialist, utilities and the roadway owner along with the first-alarm assignment.
• The incident commander or training officer adds one item to the agenda of the next regular command staff meeting asking who is authorized to declare the shift from rescue to recovery at a collapse, and what documented basis that decision requires.
• The public information officer adds a rule to the existing crisis communications plan that any early casualty estimate is released only as an estimate with its basis stated, and checks the last after-action report for any figure released without a source and time.

Takeaways

• The Loma Prieta earthquake struck at 5:04 p.m. on 17 October 1989, and the USGS gives its magnitude as 6.9, although many reports at the time gave 7.1 on a different scale.
• The USGS gives the death toll as 63, and 42 of those deaths are most commonly attributed to the collapse of the Cypress Street Viaduct in West Oakland, about 60 miles from the epicenter.
• The collapsed portion of the viaduct stood on soft bay mud and fill that amplified shaking, while a portion on firmer soil stayed up, and the non-ductile 1950s joints between the lower and upper frames failed under that shaking.
• The claim that the World Series crowd was trapped on the Cypress structure is a myth, and the real effect usually attributed to the game is lighter evening traffic because many people had gone home early to watch it.
• Early casualty estimates for the Cypress structure, built partly on assumed rush hour volume, were far higher than the final count and were revised down over several days.
• West Oakland residents and nearby workers reached trapped motorists before organized rescue was established along most of the collapse, working on a structure that nobody had assessed.
• Loma Prieta pushed California’s investment in urban search and rescue, though FEMA’s national system had roots in Mexico City 1985 and Armenia 1988, and California’s SEMS came from the 1991 Oakland Hills fire.
• The Housner board’s report led to an accelerated Caltrans retrofit program built around column jacketing, and the Cypress freeway was rebuilt on a new route around West Oakland, opening in 1997.