Cycling training on rotating shifts — what actually works

Every training plan assumes the same week, every week. A rotating roster changes three times a month. This is how I made a structured, power-based cycling plan survive one — the five rules it runs on, a worked example week for each shift type, what it costs, and how to build your own in an evening.

I am 45. I work rotating shifts: early, afternoon and night, in runs of three to five, handed to me months in advance but never repeating. In January I weighed 98 kg. In July I rode 100 km in one go, badly, and decided to do it properly. Everything below is what I actually do — you can watch it happen on the plan, day by day, including the weeks that go wrong.

Why does a normal training plan fall apart on shift work?

Open any 16-week plan and you will find Tuesday intervals, Thursday threshold, Saturday long ride. That structure quietly assumes three things a shift worker does not have: the same free hours each week, a stable sleep schedule, and recovery that resets overnight.

On a run of nights I sleep in the day, badly, and lose two to three hours a night against my normal. By the third night the deficit is measurable: resting heart rate up, overnight HRV down, and the same wattage feels about ten percent harder. If the plan says "threshold intervals Thursday" and Thursday is my third night shift, I have three bad options — ride it and dig a hole, skip it and break the plan, or ride it badly and lie to myself about it.

Most people pick the third, then quit in week six and conclude they have no discipline. The problem was never discipline. The plan was wrong.

Rule 1 — the shift decides the session, not the calendar

I stopped thinking in weekdays. Every day in my plan is labelled by shift first, and the shift determines what is even possible.

What can you train on an early shift?

Early shift is 05:30 to 14:00, so the alarm goes at 04:15. I am home by 14:30 and useless. The ride is a short evening session, 45 to 65 minutes, around 17:00, after food and a sit-down: endurance, or a short sweet spot block if the week needs one. Nothing that requires a long warm-up and a clear head. The limiting factor is not the legs, it is that I have been awake since four.

Why is the afternoon shift the best training window?

Because a 13:30 start leaves a free morning after a full night of sleep, and almost nobody uses it. I ride 08:00 to 10:30 with nothing after it but a shower and a commute. Every hard session I can move here, I move here — thresholds, over-unders, VO₂max, the long ride. Two of these mornings in a week is a productive week, and a roster with a run of afternoons is the block where fitness actually moves.

Can you ride on night shifts at all?

Yes, but only easy. Night shift is 21:30 to 06:00. I sleep through the day and ride around 16:00, forty to sixty minutes, conversational Z2 only, and I stop if it stops feeling easy. Never any intensity on a night shift. Not once in sixteen weeks. That session exists to keep the habit and move some blood, not to build anything. Treating a night-shift ride as training is the single most common way shift workers wreck a block.

What about the first night of a run?

The first night is a gift and most people waste it sleeping in. I do not start until 21:30, so the whole day is free — it takes a full session, often the long ride of the week, as long as I finish three or four hours before the shift and eat properly afterwards.

What about the day after the last night shift?

Always off. No exceptions, no "just an easy spin". That day is for sleep. Training on it costs more than it gives, every single time, and it is the day I most often want to ride because I feel guilty about the days before.

Write those windows down before you write a single interval. If a session does not fit inside a real window, it is not a session, it is a wish. My plan is generated with a hard limit per shift type, and a workout that exceeds its window fails to build at all. That one constraint removed more failed weeks than anything else I did.

What does a real week look like on each shift?

Three worked examples, all at a starting FTP of 240 W. Adjust the watts to your own — everything in my plan is written as a percentage, precisely so this transfers.

A week of afternoon shifts — the productive one

A week of early shifts — the maintenance one

A week with nights — the damage-limitation one

Rule 2 — the morning numbers get a veto

What should make you skip a session?

Every morning, before anything, I check four things: Body Battery, hours slept, overnight HRV against my own 7-day average, and how I honestly feel on a 1–5 scale. Any one of these trips the switch:

• Body Battery under 65
• Less than 6 hours of sleep
• HRV below 85% of my recent average
• Feeling rated 2 or worse

When one trips, the hard session does not happen today. And during a run of nights there is a harder rule: under 5 hours of daytime sleep means no riding at all, not even easy. On four hours of sleep a Z2 ride is not recovery, it is another stressor on a body that is already losing.

This felt like cheating when I started. It is not. The numbers are not an excuse generator — they are a lie detector, and they work in both directions. On a good morning they tell me to go hard when I would have talked myself into going easy, and that happens as often as the reverse.

What if you don't have a watch that measures HRV?

Take your resting heart rate first thing, in bed, before you stand up, for two weeks. That gives you a baseline. Five beats above it means the same thing a low HRV reading means. Add how many hours you slept and an honest 1–5 for how you feel, and you have 80% of the signal for the price of a watch you already own — or a finger and a clock.

Rule 3 — move the session, don't delete it

Here is where most adaptive plans get it wrong: they skip the workout and carry on as if nothing happened. Mine moves it. A vetoed hard session goes to the next day that can actually take it, within four days — in practice usually the next afternoon-shift morning. If no such day exists inside that window, it is gone and I do not make it up.

The distinction matters because a training block is a dose, not a schedule. Three threshold sessions in nine days is the same stimulus whether they land on Tuesday, Thursday and Saturday or on Wednesday, Sunday and the following Tuesday. What ruins a block is not moving work around — it is quietly doing less of it every week and not noticing.

The four-day limit matters too. Without it, "I'll do it later" becomes a rolling debt you never repay, and the week's structure dissolves into whatever you felt like doing.

Rule 4 — rate every session, and let it change the watts

How do you know if the intensity is right?

After every ride I score it 1 to 5. One means very easy, three means about right, five means too hard. It takes two seconds. Then the plan does something with it:

• Two "too hard" ratings in a row → the work intervals drop by 3%.
• A "very easy" → they go up by 2%.
• The correction is capped at −6% / +4%, so one bad night cannot wreck a block.

This is the closest thing to a coach that a spreadsheet can give you. An FTP test tells you where you were on one particular morning. The ratings tell you where you are this week, on this much sleep, after this many nights. On the plan you can see the corrections being applied — every rated session shows its score, and the sessions I have ridden also carry the morning numbers they were ridden on.

Rate honestly or do not bother. A plan fed flattering numbers gives back flattering numbers.

Rule 5 — the volume you can repeat beats the volume that looks good

How many hours a week can you actually train on shifts?

About 6 to 7 hours in my case, rather than the 10 to 12 a normal plan assumes. The ceiling is not motivation — it is what survives a month containing two runs of nights. Structure is three weeks of rising load, one week of recovery, three times over, each block ending with a ramp test.

Most of those hours are Z2: unglamorous, conversational endurance. The hard work is sweet spot (88–90% of FTP) and threshold (95–98%), because they give the most fitness per unit of recovery — which is exactly the currency a shift worker is short of. VO₂max appears rarely, only on afternoon-shift mornings, and never in a week with nights in it.

If those terms are unfamiliar, I keep a plain-language glossary — nothing on this site needs more than that list. And if you want the numbers in your own watts, the calculators convert any of this to your FTP.

What does this cost?

I want to be straight about the trade, because plans that promise everything are the ones people quit.

I train fewer hours than someone on a normal schedule. My hard weeks are less hard, because I will not stack intensity next to a run of nights. I lose four to six days of useful training a month to the roster, and no amount of cleverness gets them back. Progress is slower — my own honest projection is on the fitness page, and it flattens well before it gets anywhere impressive.

What I get instead is a plan that is still running in week 16. That turns out to be the whole game. Consistency at 80% beats brilliance that collapses in a month, and on a rotating roster brilliance always collapses.

How long before you see anything?

Honest timeline, from my own numbers and from what the plan projects:

• Weeks 1–3: nothing measurable. What changes is that the sessions stop feeling like an ambush — you learn what 88% of FTP feels like at 06:00 versus at 17:00, and that alone is worth the block.
• First re-test, week 4–5: the biggest single jump you will ever see, and most of it is not fitness. It is pacing, familiarity with the test, and a body that has stopped being surprised. Do not extrapolate from it.
• Weeks 6–12: the real curve. Slow, unglamorous, two or three watts a month if the roster cooperates. This is where people quit, because the graph stops being exciting exactly when the work starts counting.
• W/kg moves faster than watts. Worth knowing, because it is the number you actually feel on a climb: losing 5 kg does more for it in twelve weeks than any interval session will.

The single most useful thing I did was write the projection down before starting, in public, with a cautious and an ambitious line. Now every test either lands inside it or does not, and I cannot quietly revise the story afterwards. If you do nothing else from this page, do that.

How do you build your own version tonight?

1. Write down your real windows. For each shift type, the hours you could actually be on the bike — after commute, after food, before sleep. Be pessimistic. That is your maximum session length, and everything else follows from it.
2. Get an FTP number. Ride a ramp test. Everything becomes a percentage of it, so one number keeps the whole plan honest and a re-test every four to six weeks updates all of it at once.
3. Put the hard sessions where the good window is. For me that is afternoon-shift mornings. Find yours and defend it against everything else in your life.
4. Fill the rest with Z2 and accept that it is meant to be boring.
5. Decide your veto rules now, in writing, while you are rested. Deciding at 05:00 on four hours' sleep whether today counts as "too tired" is a decision you will get wrong.
6. Score every session and act on the score. Two hard ones in a row means the numbers are too high for the sleep you are getting, not that you are weak.

If you want a starting point rather than a blank page, my whole plan is free: all 89 sessions as Zwift files, written in percentages of FTP so they scale to your numbers without editing. No email, no sign-up.

One honest caveat

I am a shift worker who got obsessive about this, not a coach and not a doctor. These are the rules I built for my own roster and my own body, published so that someone in the same situation does not start from zero. Sleep loss from night work is a genuine health issue, not just a training inconvenience — if you are struggling with it, that is a conversation for a doctor, not a training plan.

What I can offer is the evidence. Everything here is running live: the plan as written, the morning numbers behind every decision, and the fitness curve with the real test results plotted on it. In a few months it will be obvious whether any of this worked. If you also train around shifts, tell me what you do differently — I answer everything.

The short version: the rules themselves