California — from the Salton Sea to Cape Mendocino · Right-lateral strike-slip
The San Andreas Fault is the principal boundary between the Pacific Plate and the North American Plate, and the most recognized — and most studied — earthquake fault in the United States. It runs roughly 1,200 km the full length of California, from the Salton Sea in the south to Cape Mendocino in the north, passing through or near the San Francisco Bay Area, the Los Angeles region, and the Inland Empire. The Pacific and North American plates grind past each other here at about 50 mm per year, distributed across the San Andreas and its parallel faults.
On January 9, 1857, the southern and central San Andreas ruptured in the Mw 7.9 Fort Tejon earthquake, breaking roughly 360 km (225 mi) from near Parkfield to near Wrightwood with up to about 9 meters (30 ft) of right-lateral offset. It remains the benchmark for what a full southern-segment rupture looks like. The southernmost Coachella segment last broke around 1690 — more than 330 years ago.
On April 18, 1906, the northern San Andreas ruptured in the M7.9 San Francisco earthquake, breaking about 477 km (296 mi) from San Juan Bautista to Cape Mendocino. More than 3,000 people died (the long-cited figure of ~700 is now understood to be a 3–4× underestimate) and roughly 225,000 were left homeless. It is the defining earthquake in California history and the reason the state takes seismic hazard seriously.
The fault behaves very differently along its length. The northern segment (San Juan Bautista to Cape Mendocino) is locked and accumulating strain; it last broke in 1906. The central segment (~150 km from Parkfield to Hollister) creeps continuously — slipping steadily without building up strain for large quakes, famously offseting roads, curbs, and fences. The southern segment (Parkfield to the Salton Sea) is fully locked and critically stressed; its central portion last ruptured in 1857 and the Coachella section around 1690.
The locked southern segment is the greatest concern. The Coachella section has not ruptured in over 330 years, well past its estimated 200–300 year average recurrence. A 2026 study (Burkhard et al., JGR: Solid Earth) found stress on the southern San Andreas at its highest level in a 1,000-year model and identified Cajon Pass as an "earthquake gate" that could let a future rupture jump between the San Andreas and the San Jacinto fault — potentially a larger multi-fault event.
The 1989 Loma Prieta earthquake (M6.9) struck the Santa Cruz Mountains segment, rupturing the southernmost ~48 km of the 1906 break; 63 people died and damage ran to about $6 billion. At Parkfield, the fault produces a remarkably regular series of M~6 earthquakes (1857, 1881, 1901, 1922, 1934, 1966, 2004) — the 2004 event arrived about 11 years "late" in the roughly 22-year cycle and was captured by dense instrumentation. Note: the 1992 Landers earthquake (M7.3) was not on the San Andreas — it ruptured faults in the Eastern California Shear Zone.
UCERF3 estimates roughly a 60% probability of M6.7+ shaking in Southern California and 72% in Northern California within 30 years. The ShakeOut scenario — a M7.8 on the southern San Andreas — projects more than 1,800 deaths, ~50,000 injuries, and ~$200 billion in losses. Recurrence intervals are paleoseismic averages with wide scatter, not clocks: clustering and irregularity are the norm. If you live near the fault, see our seismic risk lookup for your address.