The San Andreas Fault Explained
By Quake Tracker Live Editorial · Published November 9, 2026
The San Andreas Fault is the main boundary between the Pacific Plate and the North American Plate. It runs through California from the Salton Sea in the south to Cape Mendocino in the north, roughly 800 miles (about 1,300 km) according to the USGS. The two plates slide past each other horizontally, which is why it is called a strike-slip or transform fault, and the fault has produced some of the largest earthquakes in the history of the United States.
Our overview of U.S. earthquake zones puts California in context. This guide goes deeper into the fault itself: its geometry, its speed, the difference between its creeping and locked sections, and what is known about the next big earthquake.
Where the fault runs
In the south, the San Andreas starts near the Salton Sea in the Imperial Valley, then crosses the desert east of Los Angeles and bends through the Transverse Ranges, where it passes close to the mountains north of San Bernardino and Los Angeles County. It continues northwest through the Carrizo Plain, the Parkfield area and the hills east of Monterey Bay, then follows the San Francisco Peninsula. Offshore at Daly City, it passes just west of San Francisco, comes back on land at Bolinas and Point Reyes, and goes out to sea again at Point Arena before reaching Cape Mendocino, where it meets the Cascadia subduction zone (see The Cascadia Subduction Zone explained).
The San Andreas is not one clean crack. It is the main strand of a wider fault system that includes the Hayward, Calaveras, San Jacinto, Elsinore and many more. Together they share the motion between the plates. Following the live map of California earthquakes, you will see that most small events cluster along these branches rather than exactly on the San Andreas.
How fast does it move?
The USGS puts the long-term motion of the Pacific Plate relative to North America at about 46 mm per year (roughly 2 inches), moving northwest. The San Andreas itself does not carry all of that. Geologic studies cited by the USGS find slip rates on the fault of about 34 mm per year in central California, dropping to roughly 20 mm per year in northern California, and about 35 mm per year on the Mojave section falling to about 20 mm per year in the Coachella Valley in the south. The rest of the motion is taken up by neighboring faults and by the Eastern California Shear Zone.
That is about the speed at which a fingernail grows. It sounds slow, but the motion adds up: after 100 years, the two sides of a locked fault have been pulled about 3 to 3.5 meters (10 to 11 feet) out of alignment. When the fault finally slips, it releases that stored strain in seconds.
Creeping and locked sections
The fault does not behave the same way along its whole length, and that is one of the most useful things to know about it.
- The creeping section between San Juan Bautista and Parkfield slides slowly and continuously without large earthquakes. It produces many small ones, mostly magnitude 5 and smaller, according to the USGS.
- Locked sections to the north and south stay stuck for decades or centuries, building up strain, then break in large earthquakes.
- Parkfield, at the southern end of the creeping zone, has had six similar magnitude 6 earthquakes since 1857, about one every 22 years, which made it a natural laboratory for earthquake research.
The result is a fault that is partly quiet and partly dangerous. Visible signs of creep, such as curbs and walls in Hollister that are slowly offset, do not mean a nearby section is safe. They mean that one section is releasing strain gradually.
The big ones: 1857 and 1906
Two earthquakes define the San Andreas in the historical record, and they broke different halves of the fault.
| Earthquake | Part of the fault | Magnitude | Notes |
|---|---|---|---|
| January 9, 1857 Fort Tejon | South of Parkfield, in central and southern California | About 7.9 | Occurred in a sparsely populated California; the USGS places it south of the creeping section |
| April 18, 1906 San Francisco | Northern San Andreas, from San Juan Bautista to Cape Mendocino | About 7.9 | Rupture 296 miles (477 km) long; horizontal slip from 2 to 32 feet (0.5 to 9.7 m); felt over about 200,000 square miles |
| October 17, 1989 Loma Prieta | Santa Cruz Mountains, near the 1906 rupture | 6.9 | Rupture only about 25 miles long |
Magnitudes and rupture details from USGS pages on the 1906 earthquake and the Parkfield region. The 1857 and 1906 magnitudes are estimates.
The 1906 earthquake began at 5:12 a.m. local time with a foreshock, and the main rupture started about 20 to 25 seconds later. Strong shaking in the Bay Area lasted roughly 45 to 60 seconds. Much of the destruction in San Francisco came from fires that followed the shaking. For this and other major events, see the largest earthquakes in U.S. history.
Logarithmic scale: each gridline is 10 times the previous one.
View the data
| Earthquake | Energy released (vs. an M6.0) |
|---|---|
| Fort Tejon (1857) | M7.9 · about 710× |
| San Francisco (1906) | M7.9 · about 710× |
| Loma Prieta (1989) | M6.9 · about 22× |
| Parkfield (typical, M~6) | M6.0 · 1× |
Energy scales by a factor of about 31.6 per magnitude unit. Historic magnitudes are estimates; Parkfield events are "magnitude about 6".
What do forecasts say about the next one?
Faults do not run on a schedule, so seismologists avoid the word "overdue". What they offer are probabilities. The USGS and the Southern California Earthquake Center, in the third Uniform California Earthquake Rupture Forecast (UCERF3, 2015), estimate about a 7 percent chance of a magnitude 8 or larger earthquake somewhere in California in 30 years, and name the southern San Andreas as the fault most likely to host one. Across the whole state, a magnitude 6.7 or larger earthquake is more than 99 percent likely in that period, but that includes many faults besides the San Andreas.
The reason the southern San Andreas gets so much attention is simple: it has been quiet for a long time. Large stretches have not ruptured since 1857 or even earlier, so a lot of strain has accumulated, and the fault runs near populated areas in the Inland Empire, the Coachella Valley and the Los Angeles basin. Even so, a USGS probability is an estimate, not a prediction, and nobody can say whether a rupture will begin this year or in several decades.
How the fault affects daily life in California
- Building codes in California require newer buildings to resist strong shaking, and older concrete and wood-frame buildings are being retrofitted.
- Lifelines such as water, gas, power and highways cross the fault in several places, which is why utilities plan for fault offset.
- Preparedness matters more than prediction. Our preparedness checklist lists what to have ready, and what to do during an earthquake explains Drop, Cover and Hold On.
You can watch the ongoing small earthquakes along the fault on our homepage map, the today page, or the California page. Most of the activity you see on an ordinary day is magnitude 2 or 3 and too small to feel.
Frequently asked questions
How long is the San Andreas Fault?
The USGS describes it as about 800 miles (roughly 1,300 km) long, from the Salton Sea in Southern California to Cape Mendocino in the north. Other published estimates for the main fault zone are slightly shorter, around 1,100 km.
How fast does the San Andreas Fault move?
The Pacific Plate moves northwest relative to North America at about 46 mm (roughly 2 inches) per year, and geologic studies put the San Andreas itself at about 20 to 35 mm per year depending on the section. The rest of the motion is shared by neighboring faults.
Will California fall into the ocean?
No. The fault is a horizontal slide, not a gap. The Pacific Plate is moving northwest past North America, so in many millions of years the land west of the fault will be farther north, not underwater.
What was the 1906 San Francisco earthquake?
A magnitude 7.9 earthquake on April 18, 1906 that ruptured 296 miles (477 km) of the northern San Andreas Fault. Strong shaking lasted about 45 to 60 seconds in the Bay Area, and the fires that followed caused much of the damage.
Is the San Andreas Fault overdue?
Seismologists do not use that word, because faults do not follow a timetable. The southern San Andreas has not had a major rupture since at least 1857, which means a lot of strain has built up, and the USGS gives probabilities over decades instead of dates.
Sources and further reading
- USGS: The Northern California Earthquake of April 18, 1906
- USGS: Parkfield earthquake experiment, geology
- USGS: Will California eventually fall into the ocean?
- USGS: Earthquake facts and earthquake fantasy
- Southern California Earthquake Center: UCERF3
This guide is general information, not professional or emergency advice. Quake Tracker Live is not an official warning system; follow instructions from local authorities.