San Andreas Fault Map

The San Andreas fault line runs about 1,200 km (750 miles) through California, from the Salton Sea in the south to Cape Mendocino in the north; the map above traces it from the USGS Quaternary Fault Database. It is the main boundary between the Pacific Plate and the North American Plate. The Pacific Plate, on the west side, moves northwest relative to the North American Plate at about 46 mm (1.8 inches) a year, and the San Andreas and its neighbouring faults share that motion. Measured on the mapped fault zone, it passes about 8 miles from central San Francisco, about 33 miles from downtown Los Angeles, and within a few miles of San Bernardino, Palmdale, and Palm Springs.

Where Is the San Andreas Fault?

From the triple junction at Cape Mendocino, where three plates meet, the fault runs south near the coast and passes just west of San Francisco. Tomales Bay and Crystal Springs Reservoir lie in the straight trough that marks the fault zone. It continues down the Santa Cruz Mountains to San Juan Bautista, through the creeping section to Parkfield, across the Carrizo Plain, past Frazier Park, through Tejon Pass, and past Palmdale and Wrightwood to Cajon Pass near San Bernardino.

South-east of Cajon Pass the fault stops being one clean break. Several branching faults, including the San Jacinto and Banning faults, share the plate motion, and the name “San Andreas” is generally given to the north-easternmost branch, which runs down the Coachella Valley past Indio toward the Salton Sea. Published lengths differ with where those ends are drawn: about 1,200 km (Southern California Earthquake Data Center) to about 1,300 km (USGS).

Straight-line distance from each city’s center to the nearest mapped strand of the San Andreas fault zone, measured on the USGS trace shown in the map above (rounded):

Distance to the fault is only one input to shaking risk: soil, building type, and the other faults nearby matter as much. For a specific address, use the seismic risk lookup, which reports the expected ground shaking and the nearest mapped faults.

How the Fault Moves

The San Andreas is a right-lateral strike-slip fault: stand on either side and look across, and the ground on the far side has moved to your right. In 1906 that sideways jump offset roads, fences, and rows of trees; the road across the head of Tomales Bay moved almost 21 feet. Along the fault, slip averages roughly 20 to 35 mm a year. The USGS estimates that since the fault zone started moving, about 28–30 million years ago, the two sides have slid about 300–350 km (186–220 miles) past each other.

The fault does not behave the same way along its length. The northern section, north of San Juan Bautista, is mostly locked: it produced the 1906 earthquake and has had few earthquakes since, the M6.9 Loma Prieta earthquake of 1989 being the notable exception. The central creeping section, from San Juan Bautista to Parkfield, slips steadily without locking, producing many small earthquakes (mostly M5 and smaller) but no large ones in the historical record. Parkfield is the transition. The southern section, south of Parkfield, is locked again and produced the 1857 earthquake.

1857 Fort Tejon

On January 9, 1857, the San Andreas ruptured from near Parkfield almost to Wrightwood in the Mw 7.9 Fort Tejon earthquake, a surface rupture of about 225 miles (360 km). The ground was offset by as much as about 9 meters (30 ft) on the Carrizo Plain. One death was recorded. It is the last “Big One” on the southern San Andreas.

1906 San Francisco

On April 18, 1906, the northernmost 296 miles (477 km) of the fault broke, from northwest of San Juan Bautista to the triple junction at Cape Mendocino, in the San Francisco earthquake, given by the USGS as magnitude 7.9 (older moment-magnitude estimates put it near 7.7). Shaking and the three-day fire that followed killed an estimated 3,000 or more people (the long-quoted figure of about 700 is now regarded as a large undercount) and left about 225,000 of San Francisco’s roughly 400,000 residents homeless.

Loma Prieta and Parkfield

The M6.9 Loma Prieta earthquake of October 17, 1989, slipped on the San Andreas in the Santa Cruz Mountains, about 100 km south of Oakland and San Francisco. It killed 63 people and caused an estimated $6 billion in property damage. It barely broke the surface: geodetic data imply about 6 feet of sideways slip and 4 feet of reverse slip on the fault, while the only cracks that may mark the fault at the surface showed about 2 centimeters of offset.

At Parkfield, moderate earthquakes struck at fairly regular intervals in 1857, 1881, 1901, 1922, 1934, and 1966, an average of about 22 years apart. The USGS expected the next one between 1988 and 1993; it came on September 28, 2004, as a Mw 6.0. Even the most regular stretch of the fault did not keep to a schedule.

When Is the Next Big One?

At many places along the San Andreas, more time has passed than the average gap between large earthquakes. On the southernmost stretch, from Palm Springs to the Salton Sea, large earthquakes come about every 200–300 years, and the most recent was about 300 years ago. At Wrightwood, a 3,000-year record of 29 surface-rupturing earthquakes gives an average gap of about 100 years; only four past gaps there were longer than the current one, which began in 1857. The USGS cautions that “overdue” is more complicated than it sounds, because gaps vary widely.

The official forecast, UCERF3 (USGS, California Geological Survey and SCEC, 2015), gives these chances of a magnitude 6.7 or larger earthquake in the 30 years from 2014 (fault figures are for a point on the fault):

UCERF3 ranks the southern San Andreas as the fault most likely to host a large California earthquake. The northern section’s lower number partly reflects how recently it broke in 1906, and in the Bay Area the Hayward–Rodgers Creek and Calaveras faults currently rival or exceed it. A 2026 study (Burkhard, Smith-Konter, Scharer & Sandwell, JGR: Solid Earth) modeled present-day stress around Cajon Pass; according to the authors’ University of Bern release, stress on the San Jacinto–Bernardino segment is above anything in the 1,000-year simulation, and the neighbouring Mojave South segment of the San Andreas is similarly stressed. The authors say plainly that this is not a prediction of when an earthquake will happen. The USGS ShakeOut scenario, a magnitude 7.8 on the southern San Andreas, models about 1,800 deaths and $213 billion in economic losses.

Common Questions

Will California fall into the ocean? No. The San Andreas moves mainly sideways, and California spans two plates with nowhere to fall. As the USGS puts it, Los Angeles and San Francisco will instead one day sit next to each other.

Is the San Andreas the only fault to worry about? No. The region-wide odds above are far higher than the San Andreas-only figures because they count every other fault too. The 1994 Northridge earthquake (M6.7) occurred on a previously unrecognized fault, and the 1992 Landers earthquake (M7.3) broke faults in the Mojave Desert, not the San Andreas. The USGS notes that the next very damaging earthquake could easily be on another fault.

Can you see the fault? In many places, yes: from the air it shows as a straight trough lined with lakes, bays and valleys, and on the ground as long straight escarpments, small sag ponds, and stream channels that jog sharply to the right where they cross it. The Carrizo Plain, where the 1857 earthquake offset the ground by as much as about 9 meters, is a good place to see it.