Understanding Time in San Francisco: What Actually Happens Here
Pacific Time runs on UTC-8 during standard time and UTC-7 when daylight saving is active. That sounds straightforward until you try to schedule a call with someone in New York during the transition weeks, or your server logs start showing an hour you didn't account for. I've spent years working with systems that touch multiple time zones, and San Francisco's particular setup causes more headaches than most people expect. The Pacific Time Zone covers the West Coast, but it's not just about switching clocks back and forth. California observes daylight saving time from the second Sunday in March through the first Sunday in November. Arizona and most of Hawaii don't, which creates that awkward annual period where the time difference between San Francisco and Phoenix flips by an hour. I once had a deployment script fail because it was comparing timestamps across zones during one of these shift windows, and it took me three hours to trace the bug back to the DST boundary. The fix was simply to always store and compare everything in UTC internally, then convert to local time only at the presentation layer. Here's what most people miss about Pacific Time. The transition dates aren't fixed calendar dates—they're tied to Sundays. That means the exact moment of the switch changes every year, ranging from March 8th to March 14th for the spring forward, and November 1st through November 7th for the fall back. If you're building scheduling software or running automated jobs that depend on precise timing, hardcoding these dates will break you within a year. Use the IANA timezone database instead, specifically the identifier America/Los_Angeles. Never use PST or PDT as a timezone identifier in code—those are ambiguous. PST technically refers only to standard time, and some systems will reject it or fall back to treating it as UTC-8 year-round, which is wrong half the year.
Another thing nobody warns you about is the gap and the fold. When springs forward in March, you lose an hour—2:00 AM becomes 3:00 AM instantly. Any code that assumes every day has exactly 86,400 seconds will silently skip that hour. When it falls back in November, you gain an hour—2:00 AM happens twice. This is where most scheduling bugs hide. A job set to run at 2:30 AM will execute twice on that day unless you explicitly handle the ambiguity. I've seen payment processing systems charge customers twice because of this exact scenario during a fall-back night. San Francisco sits at approximately 37.77 degrees north latitude, which puts it squarely in the Pacific timezone rather than anywhere near the Alaska or Hawaii-Aleutian zones. The continental US has four primary time zones, and the borders between them don't follow state lines cleanly. Eastern Idaho is on Mountain Time despite being in a state that's mostly Mountain. portions of Kentucky and Tennessee straddle Eastern and Central. But within California itself, it's uniform—no exceptions, no half-counties on a different schedule like you get elsewhere. If you need to convert times reliably, the approach that actually works is to treat UTC as your source of truth. Parse incoming timestamps into UTC immediately. Do all your arithmetic in UTC. Convert to Pacific Time only when displaying to a user. Libraries like Moment-Timezone for JavaScript, pytz or zoneinfo for Python, and the built-in LocalDateTime classes in Java all handle the IANA database correctly if you use them properly. The most common mistake is calling .getTimezoneOffset() in JavaScript and assuming it returns a constant value—it doesn't, it changes twice a year.
For anyone running servers or cron jobs that depend on San Francisco time, configure your system timezone explicitly. Docker containers especially are notorious for defaulting to UTC and then confusing everyone when logs show one time but the application thinks it's another. Set the TZ environment variable to America/Los_Angeles in your container or host configuration, and verify it with a simple date command. Takes thirty seconds and prevents weeks of debugging later. The biggest practical limitation of relying on any timezone-aware system is that historical timezone data can shift. Government bodies change daylight saving rules occasionally. China abolished its DST program in 1991, and while that doesn't affect San Francisco directly, it's a reminder that the IANA database contains historical transitions that older software versions may not know about. If you're dealing with logs or records from before your software's timezone library was last updated, those timestamps could be wrong. Keeping your timezone data current isn't optional if accuracy matters.
Get the Full Details
