When Does Fall Start: Two Clocks, One Season, and Why the Dates Disagree

Fall starts on two different dates, depending on which clock you use. Astronomical fall begins at the September equinox, which falls on September 22 in 2026 (September 21 in some years and time zones). Meteorological fall starts on September 1 every year, by convention. Neither date is "wrong"; they answer different questions, and the gap between them is about three weeks.

The disagreement is not trivia. It shapes how climate scientists report seasonal data, how farmers and gardeners plan, and why the first crisp morning often arrives weeks before either official date. Understanding both clocks, and the physics behind them, explains why fall feels like it starts at different times to different people.

The two definitions of fall

Astronomical seasons are defined by Earth's orientation relative to the Sun. Fall begins at the autumnal equinox, the moment the Sun crosses the celestial equator heading south. Day and night are then nearly equal everywhere on Earth, and the Northern Hemisphere begins losing daylight faster than at any other time of year.

Meteorological seasons are defined by calendar months, not orbital geometry. Fall runs September 1 through November 30 in the Northern Hemisphere. This convention exists because climate and weather records are compiled in whole months, and month boundaries make seasonal statistics far easier to compute and compare.

The two definitions serve different masters. Astronomy tracks the planet's geometry. Meteorology tracks human record-keeping and the annual temperature cycle. Both are legitimate; they simply measure different things.

Why the equinox is not a fixed date

Earth's orbit is an ellipse, and its axis is tilted about 23.4 degrees. That tilt drives the seasons: it changes how directly sunlight strikes each hemisphere and how long the Sun stays above the horizon. The equinox marks the instant the tilt points neither toward nor away from the Sun.

That instant does not land on the same calendar date every year. A tropical year, the time between successive equinoxes, is about 365.24 days. Our calendar year is 365 days, plus a leap day roughly every four years. The mismatch shifts the equinox by about six hours earlier each common year, then jumps back after a leap year.

The result: the September equinox drifts across September 21, 22, and occasionally 23. Over the coming years, it falls on September 22 in 2026, September 23 in 2028, and September 22 again in 2029. The exact moment also depends on your time zone, because the equinox is a single global instant expressed in local time.

There is a deeper wrinkle. The equinox is often described as the moment day and night are equal, but that is not quite true. Atmospheric refraction bends sunlight around the curve of Earth, and the Sun's disc has width, so "equal day and night" arrives a few days before the astronomical equinox at most latitudes. The technical term is the equilux.

How daylight actually drains away

Fall's most reliable signal is not temperature but light. After the September equinox, the Northern Hemisphere tilts away from the Sun, and the rate of daylight loss accelerates. At mid-northern latitudes, the day shortens by roughly two to three minutes per day around the equinox, faster than at any other time of year.

That loss compounds. By the November solstice window, many mid-latitude locations have lost three to four hours of daylight compared with late June. The change is gradual enough to miss day to day but obvious across a month. This is the mechanism behind the seasonal mood shifts many people notice as fall deepens.

Temperature lags light. The ground and oceans store summer heat and release it slowly, so the coldest weeks arrive well after the daylight minimum. That lag is why meteorological fall (September through November) captures the cooling season more cleanly than the astronomical window does.

Timeline: fall's arrival across markers

The lesson from the table: "fall" arrives at different markers at different times, and none of them is universal. A coastal city may see its first frost weeks after an inland valley at the same latitude.

Comparison: which fall definition fits which purpose

No single definition wins. The right one depends on what you are trying to measure or plan.

Edge cases: latitude, altitude, and the tropics

Fall behaves very differently depending on where you stand. Near the equator, there is essentially no autumn: daylight barely changes, and temperatures stay flat year-round. The seasonal signal strengthens as you move toward the poles.

At high latitudes, fall arrives abruptly. Parts of Alaska, northern Canada, and Scandinavia lose daylight rapidly in September and October. Some Arctic locations see the Sun set for the season in late November, ending fall with the polar night rather than a gradual cooling.

Altitude matters as much as latitude. A mountain town at 7,000 feet can see frost and early leaf change while a valley 50 miles away stays green for weeks. Microclimate, slope aspect, and proximity to water all shift the effective start of fall.

The Southern Hemisphere runs the opposite calendar. Its fall begins in March, at the March equinox, and its meteorological fall runs March through May. Any global discussion of "when fall starts" must specify the hemisphere.

A counterintuitive caveat worth flagging: the equinox does not guarantee cool weather, and the meteorological start does not guarantee summer heat is over. In many years, September brings both the hottest days of the year and the first frost within the same month, depending on location and weather pattern.

Misconceptions worth correcting

Myth: fall always starts on September 22. The equinox drifts across three dates and depends on time zone. Myth: day and night are exactly equal on the equinox. Refraction and the Sun's width push true equality a few days earlier. Myth: meteorological fall is a "fake" season. It is the standard basis for climate records and forecasts. Myth: leaves turn because of cooler weather alone. Daylength and species genetics drive the change; temperature modulates timing. Myth: fall starts at the same moment everywhere. The equinox instant is global, but local dates, seasons, and hemisphere reverse the calendar.

Practical actions: planning around fall's real start

Use the right clock for the right task, and build in a buffer for local variation.

For frost-sensitive plants, ignore both calendars and look up your local average first-frost date; plant or protect accordingly. For daylight and mood planning, track the astronomical date and your latitude, because light loss, not temperature, drives seasonal effects. For climate or energy records, use meteorological months so your data lines up with standard reporting. For travel and foliage, check regional peak-color forecasts, which reflect species and weather, not the equinox. For outdoor work and safety, note that earlier sunsets arrive weeks before the calendar says fall, so adjust lighting and schedules in September. Red flag: do not assume a warm September means a mild fall; early-season warmth says little about the weeks ahead.

Limitations and what the dates cannot tell you

Neither fall start date predicts weather. The equinox is a geometric instant with no bearing on temperature, precipitation, or storm activity. The meteorological date is a bookkeeping convention, not a forecast.

Local conditions, elevation, ocean influence, and year-to-year weather patterns all shift when fall actually feels like fall. Any single date is a framework, not a guarantee. The most reliable guide to your fall is your own local records and frost data.

The dates also drift slowly over centuries because of small changes in Earth's orbit and rotation, though the shift is negligible across a human lifetime. For practical purposes, the September 21–23 window and the September 1 convention will remain stable.

If you are using fall timing to plan anything with health, safety, or financial consequences, treat these dates as general scheduling anchors only. Consult local experts and official forecasts for decisions that carry real risk.