The Clock Change, Measured: How America's Wake-Up Time Moves the Week After Fall Back

The week before and the week after, side by side

Here is the headline number. On weekdays in the week before fall back, the average American in the survey woke at 6:59 a.m. In the week after, the average was 6:55 a.m. That is a difference of 4.8 minutes earlier.

Before you file that away as "the clock change makes people wake five minutes early," it helps to look at how sure that number is. When we resampled the data 2,000 times to see how much the estimate wobbles, the plausible range ran from about 12 minutes earlier to about 3 minutes later. Zero sits comfortably inside that range. In plain terms, the data cannot tell a true shift apart from no shift at all.

For this measurement, a wake time is the moment the first stretch of sleep ends between 4 a.m. and noon. The survey codes that sleep as its own activity, labeled simply "Sleeping" in the American Time Use Survey activity lexicon, separate from time spent lying awake. A bedtime is the start of the last sleep episode that begins after 6 p.m.

What a normal fall fortnight looks like

The brief for this piece called for a day-by-day view: the seven days before the clock change, the changeover Sunday itself, and the seven days after. That is exactly what the table below shows, pooled across 22 years of diaries from the ATUS 2003–2025 multi-year data files. Day 0 is the Sunday the clocks fall back, so day −6 is the Monday before and day +1 is the Monday after.

Weekday days draw on roughly 440 to 510 diaries each; weekend days draw on 1,150 to 1,275, because the survey deliberately collects more weekend diaries.

Reading the table

The first thing your eye probably catches is the weekly rhythm. Weekday mornings cluster just before 7 a.m. Weekend mornings sit right around 8 a.m. Saturday nights run about half an hour later than the rest of the week. That pattern repeats on both sides of day 0 almost exactly, like two copies of the same week laid end to end.

The second thing is what's missing. There is no visible break at the changeover. The Monday after looks like the Monday before. The Wednesday after looks like the Wednesday before. If you covered up the day labels, it would be hard to say which week came first.

Why the median column jumps around

The median wake time hops between 6:30, 6:45 and 7:00 a.m. on weekdays, even though nothing is changing. That is a quirk of how people report time, not a quirk of their mornings. In this window, 89.8% of weekday wake times land exactly on the hour or half hour, and 96.7% land on a quarter hour. People say "I got up at 6:30," not "I got up at 6:37."

Because of that rounding, the median can only move in 15- or 30-minute steps. It is a helpful snapshot of the "typical" time, but it cannot detect a change of a few minutes. That is why every comparison in this piece uses the weighted average, which can.

Comparing ordinary weeks to each other

A shift of 4.8 minutes sounds small, but small is not the same as zero. To see whether it means anything, we asked a simple question: how much do average wake times move from one week to the next when the clocks don't change at all?

We ran the exact same comparison on neighboring weeks. The week before fall back, compared with the week before that, moved by 4.2 minutes later. The second week after fall back, compared with the first week after, moved by 3.4 minutes earlier. Both of those are weeks with no clock change, and both moved by about as much as the fall-back week did.

So the honest reading is this: the fall-back week looks like any other autumn week. A few minutes of drift up or down is simply the normal week-to-week wobble in a national survey, and the clock change doesn't push the morning outside that wobble.

Checking the days the brief asked about

The brief also asked for specific checkpoints on day +1, day +3 and day +7. Because each offset always lands on the same weekday, the fairest comparison pairs each day with the same weekday a week earlier.

The ranges on single days are wide, because each weekday only has a few hundred diaries behind it. Even so, none of them lands clearly away from zero, and none of the point estimates is larger than six minutes.

Bedtimes on both sides of the change

If mornings were going to move, you might expect evenings to move with them. They don't. On weekdays, the average bedtime was 10:41 p.m. in the week before fall back and 10:38 p.m. in the week after, a difference of 2.8 minutes, with a plausible range from about 10 minutes earlier to about 4 minutes later.

The placebo check tells the same story here. The week before fall back, compared with two weeks before, moved by 3.6 minutes earlier. Bedtimes do seem to creep a few minutes earlier as November goes on, with the second week after landing another 5 minutes or so earlier, but that looks like a slow seasonal slide rather than a step at the changeover.

The changeover Sunday itself

Day 0 is the only day in the window that might look a little different. Its average bedtime, 10:28 p.m., is about five minutes earlier than the Sunday before. But the following Sunday, day +7, comes in at 10:29 p.m., almost exactly the same. Sunday nights in November simply run a few minutes earlier than the late-October Sunday that started the window, and the changeover Sunday sits right in line with that.

Where schedules are loosest

Weekends are where you might expect the clock change to show up most clearly. There is no commute, no school bell and no 9 a.m. meeting holding the morning in place. If people's mornings were going to drift with the hour, Saturday and Sunday would be the place to see it.

They don't drift there either. Comparing the weekend before fall back (Sunday day −7 and Saturday day −1) with the weekend that starts on the changeover day (Sunday day 0 and Saturday day +6), average wake time changed by 3.3 minutes earlier and average bedtime by 5.1 minutes earlier. Both ranges include zero. The median weekend wake time was 8:00 a.m. on every single weekend day in the window.

That consistency is quietly striking. On the very first morning after the clocks fall back, when you might picture half the country waking at 7 a.m. on the new clock because it feels like 8, the typical weekend riser still reports getting up at 8.

Three age groups, the same result

The brief asked whether the picture looks different at different stages of life. We split the weekday diaries into three age bands and ran the same before-and-after comparison for each.

The 65-and-older group has the largest point estimate, a little over seven minutes earlier. It also has the fewest diaries and the widest range, and that range includes zero. Read as a group, the table says the same thing as the national number: no age band shows a shift the data can distinguish from no shift.

Same alarm, different sky

If the national schedule doesn't move on the clock, something else has to give, and that something is the sun. Daylight saving time ends by turning the clock back an hour, which means sunrise and sunset both arrive an hour earlier on the clock overnight.

Chicago makes a clean example. According to the U.S. Naval Observatory, sunrise in Chicago is at 7:22 a.m. on Saturday, October 31, 2026, and at 6:23 a.m. on Sunday, November 1. Sunset moves from 5:46 p.m. to 4:45 p.m.

Now set that next to the survey's weekday wake time of about 6:55 to 7:00 a.m. The week before the change, a typical riser in Chicago is up roughly 25 minutes before the sun. The week after, at the same clock time, that same alarm goes off roughly half an hour after sunrise. Nothing about the alarm changed. The sky did.

Why the clock wins

The data can't tell you why people stick to the clock, and this piece doesn't try to guess at anyone's motives. What it can show is the outcome: American mornings and evenings are anchored to clock times, and those anchors hold right through the changeover. Measured against the clock, the shift is close to zero. Measured against the sun, it is a full 60 minutes.

If you've read our earlier piece on fall daylight, this is the other half of that story.

Our 50-city ranking of daylight lost between the equinox and Halloween measured how the sun moves through the fall; this one measures how the clock-based schedule does not.

There is no jump to fade from

The original question behind this piece was how many days it takes for the fall-back shift to fade. The data answer it in an unexpected way: there is no measurable shift on the clock in the first place, so there is nothing to fade.

You can see that most clearly by stepping back from single days to whole weeks. Here are average weekday wake times for six consecutive weeks around the change.

All six weeks sit inside a 10-minute band, and the week of the change sits right in the middle of it. If the clock change produced a jump that faded over a week or two, you would expect to see a dip at the changeover followed by a climb back. Instead, the line wanders gently up and down the way survey averages do.

What "adjusting" can and can't mean here

Many articles about fall back describe a few days of adjustment. This data can't confirm or contradict anything about how a person feels during those days, because it doesn't measure feelings at all. It measures when people report waking and going to bed, and on that measure, the national schedule shows no adjustment period because it never leaves its usual times.

The data and the dates

Everything in this piece comes from public microdata, computed by Daily Vitality. The American Time Use Survey asks one person per household to walk through a single day of their life, and the ATUS User's Guide describes that diary as starting at 4 a.m. the previous day and ending at 4 a.m. on the interview day. That 4 a.m. start is why wake times are measured from 4 a.m. onward.

Each diary was lined up against the fall-back Sunday of its own year, and the rule for that Sunday changed partway through the window. Through 2006, clocks went back on the last Sunday in October. The Energy Policy Act of 2005 moved the end of daylight saving time to the first Sunday in November starting in 2007, which NIST describes as the first change to the rules in more than 20 years.

Current federal law, 15 U.S.C. 260a, sets the change at 2 a.m. on the first Sunday of November, which in 2026 is November 1.

Who was left out, and why

Arizona and Hawaii don't observe daylight saving time; the U.S. Department of Transportation lists Hawaii and most of Arizona among the places where it isn't observed. We planned to use residents of those two states as a comparison line. In practice, only 272 diaries from Arizona and Hawaii fall within the 15-day window across all 22 years, and only about 50 per week on weekdays. That is too few to draw a reliable line, so they were left out of every estimate here.

The window also stops at 2024. Survey data for 2025 has been released, but a federal shutdown in October and November 2025 left a gap in diary collection that covers the entire 2025 fall-back window, so 2025 contributes no diaries.

Weights, rules and robustness

Every figure is weighted using the survey's own weights, as the user guide recommends for the multi-year files, with the special 2020 weight used for that year. The plausible ranges come from resampling the diaries 2,000 times.

We also tested whether the result depended on how wake and bed times were defined.

• Counting a brief middle-of-the-night wake as part of the same night's sleep: weekday wake change of −6.1 minutes (range −13.5 to +1.3).
• Using only sleep already in progress at the 4 a.m. diary start: −6.3 minutes (range −13.5 to +0.8).
• Leaving out 2020, when data collection was disrupted in the spring: −5.9 minutes for wake time and −3.0 minutes for bedtime.
• Defining bedtime as the first evening sleep episode instead of the last: −5.2 minutes (range −12.1 to +1.7).

Every version lands within a few minutes of the main result, and every range includes zero.

What this data can't tell you

A few other limits are worth knowing:

• Times are self-reported and rounded. Most people report wake times on the hour or half hour, which is why averages, not medians, carry the comparisons.
• The extra hour isn't visible. Every diary in the window totals exactly 1,440 minutes, including the Saturday diary that runs through the 25-hour changeover night. The public files don't show that night's extra hour as extra time.
• Clock time only. This piece measures when people report waking and going to bed. It makes no claims about how well anyone slept, how they felt or how the change affected their health.

If you're curious about the longer-term picture of when Americans go to sleep, that is a different measurement with its own dataset.

Our look at how average sleep duration and bedtimes have shifted across generations follows bedtimes over decades rather than days.