The Hayward Fault is commonly called the most dangerous fault in the United States — not because it is the biggest, but because it is the single most urbanized active fault in the country. It runs along the western base of the East Bay hills, directly through densely populated Richmond, Berkeley, Oakland, San Leandro, Hayward, and Fremont, with hundreds of homes sitting on the trace and freeways, BART, and gas and water lines crossing it. It last ruptured in 1868, and it has since stored enough strain to produce another earthquake like that one. No one can say when that will happen.
The USGS hazard model splits this fault into 3 sections, each with its own geometry and a geologic slip rate. Each rate says whether it rests on site slip-rate data in the USGS earthquake-geology database or only on the fault’s mapped slip-rate class.
| Section | Movement | Slip rate | Geometry |
|---|---|---|---|
| Hayward (north) CA | Right-lateral strike-slip | 9 mm/yr (7–11)site data | dips 82° E · 0–11 km deep |
| Hayward (south) CA | Right-lateral strike-slip | 9 mm/yr (7–11)site data | dips 76° E · 0–13 km deep |
| Hayward (south, extension) CA | Right-lateral, oblique reverse | 5 mm/yrrate class only | dips 82° E · 0–5.1 km deep |
Source: USGS NSHM23 fault sections database v3 and western US geologic deformation model v1.0 (on-plane slip rates, preferred with low–high range; site data = measured sites in the NSHM23 earthquake-geology database, rate class only = the fault’s mapped slip-rate category); central US and Cascadia sources from the model’s source files (code.usgs.gov/ghsc/nshmp/nshms/nshm-conus).
The fault bisects the UC Berkeley campus (it runs through California Memorial Stadium) and the cities of the East Bay. The population near the fault grew from about 24,000 in 1868 to more than 2.4 million today; a USGS scenario puts more than 7 million people in the region that would feel strong to very strong shaking, with over 1.5 million workers and more than $100 billion a year in wages exposed. A repeat of 1868 today is projected to cause economic losses exceeding $120 billion. The "most dangerous" label reflects this concentration of people and infrastructure on the fault — its danger is probability multiplied by exposure.
On October 21, 1868, the Hayward Fault ruptured in a M6.8 earthquake (estimates range 6.8–7.0), breaking about 32 km (20 mi) of surface trace from Warm Springs (Fremont) north to San Leandro, and possibly as far as Berkeley, with up to about 2 m (6 ft) of offset. About 30 people died and strong shaking lasted more than 40 seconds. Before 1906, it was known as the "great San Francisco earthquake." Paleoseismic studies find a large earthquake roughly every 140–160 years — and it has been about 158 years since 1868.
Parts of the Hayward Fault creep steadily at the surface (about 3–6 mm/yr), visibly offsetting curbs, streets, and buildings. But that creep is confined to the topmost few kilometers and accounts for only about one-third of the fault's long-term surface slip; below that depth the fault is locked and storing strain, and parts — notably the northern section — are locked all the way to the surface. Creep does not relieve the hazard; the long-term slip rate is about 9 ± 2 mm/yr, and most of it accrues in large earthquakes.
Geologists had long accepted that the Hayward and Rodgers Creek faults connect at depth. In 2016, researchers mapped the actual fault trace linking them beneath San Pablo Bay (Watt et al., 2016), making a single ~140 km system. A rupture of the combined system could reach about M7.2–7.4 — larger than the Hayward alone (modeled at about M6.9). The USGS HayWired scenario models a M7.0 mainshock rupturing about 84 km (52 mi). For the San Francisco Bay Area, UCERF3 estimates about a 72% probability of a M6.7+ earthquake within 30 years; for California as a whole the figure is over 99%.
The Hayward Fault is the reason the East Bay takes seismic retrofitting seriously. Its paleoseismic dates carry some uncertainty (a 2014 radiocarbon recalibration revised the prehistoric timeline), and its probabilities depend on the model used — but the central fact is plain: a major fault under millions of people, right at its average recurrence interval. See the Calaveras Fault, the California page, and our seismic risk lookup.