Doing your easy sessions with low glycogen stores makes your muscles adapt harder. The Marquet 2016 trial did this for 21 trained triathletes and got a 2.9% improvement in 10-km run time — with no extra carbohydrate, no extra training volume, just different timing.
That’s the idea. Now here’s where most athletes go wrong with it.
What “Train Low, Compete High” Actually Means
Carbohydrate periodization is not a diet. It’s a scheduling strategy. You do some sessions with reduced muscle glycogen to amplify the molecular adaptation signal. You do other sessions — your intervals, your long rides, your race-pace bricks — fully fueled.
The phrase “train low” refers to muscle glycogen, not total daily carbohydrate intake. In the Marquet trial, both groups ate the same amount of carbohydrate per day: 6 grams per kilogram of bodyweight. The sleep-low group just ate it at different times, clustering carbs around hard sessions and eliminating them overnight before easy morning sessions.
Same food. Better adaptation.
This is a critical point. Train-low is not a reason to eat less. It’s a reason to eat smarter.
The Adaptation Signal: Why Low Glycogen Amplifies the Molecular Response
When muscle glycogen drops below a certain level, your body reads the situation as a stress signal and responds by building more infrastructure. Think of it like a power grid manager who, after the grid runs nearly dry, orders ten new substations to prevent it happening again. The grid doesn’t need to go completely dark — just close to empty.
The pathway works like this: low glycogen raises the AMP:ATP ratio in muscle cells. That activates AMPK (AMP-activated protein kinase), which phosphorylates PGC-1alpha, the master switch for mitochondrial growth. The result is more mitochondria, more fat-burning enzymes, and a higher ceiling for aerobic output.
In everyday terms: your muscles get better at burning fat and producing power because they spent some time running lean.
Across 32 controlled train-low studies, the molecular signal is consistent. 73% of cell signaling studies show augmented AMPK activation. 75% of gene expression studies show upregulated mitochondrial genes. 78% of enzyme activity studies show greater fat-oxidizing enzyme output.
Those are strong numbers. The catch is coming.
The Carbohydrate Periodization Signal vs. Performance Gap
Here’s the number that most train-low content skips entirely.
Only 37% of performance studies in that same dataset show actual race or time-trial improvement. The molecular machinery gets built. But for 63% of studies, it doesn’t translate to faster times.
Translation: your body builds more mitochondria, but you don’t always run faster.
Why? Two likely reasons. First, if the train-low sessions are hard enough to impair training quality, you lose more from the reduced effort than you gain from the adaptation signal. Second, if you do too many low sessions, you can’t complete your key interval work at the intensity that triggers the performance gains.
The Yeo and Hawley 2008 study showed this directly. Athletes who started their second daily session with low glycogen generated 8% less power during the high-intensity intervals. Both groups improved their time trial performance by roughly 10% over three weeks — but the low-glycogen group got there with worse-quality intervals. Push this too far and the adaptation signal can’t compensate for the quality loss.
This chart is why you need a decision rule for which sessions go low — not a blanket policy of training depleted.
The Glycogen Threshold Window
The adaptation signal peaks in a specific range. Impey et al. 2018 identified the glycogen threshold: roughly 100 to 300 mmol per kilogram dry weight of muscle. In this range, AMPK is maximally sensitive and PGC-1alpha activation is strongest.
Above 500 mmol per kilogram (a well-fed, rested muscle), the signal is blunted. Below 100 mmol (a fully depleted muscle), training quality collapses and you start cannibalizing protein for fuel.
The table below shows where your glycogen likely sits at each stage of the sleep-low protocol — and which stage hits the adaptation sweet spot.
| Glycogen State | Level (mmol/kg DW) | Adaptation Signal | What It Means Practically |
|---|---|---|---|
| Resting, well-fed athlete | ~648 | Blunted | Hard sessions start here |
| Post hard evening session | ~216 | Entering signal zone | Stop carbs now for sleep-low |
| Morning after sleep-low | Below 100 | Peak AMPK activation | Start easy session here |
| Fully depleted | Under 50 | Signal collapses; protein breakdown begins | Don’t go this low |
Source: Impey et al. 2018 (PMC5889771); glycogen conversion data from Murray & Rosenbloom, Nutrition Reviews, 2018 (PMC6019055).
The signal zone is roughly 100 to 300 mmol/kg DW — that post-evening-session level of around 216 falls right into it. A resting athlete at 648 is far above it. A fully drained athlete under 50 has blown past it.
Sleep-low works because the evening hard session lands you in the zone. The easy morning session keeps you there without pushing you past the floor.
Sleep-Low: What the Marquet Data Actually Shows
The Marquet 2016 protocol ran for three weeks with 21 trained triathletes. The sleep-low group ate carbohydrates before and during their evening hard session, then consumed zero overnight carbs, then completed an easy morning session before eating. The control group spread their carbs more evenly, including overnight.
The sleep-low group saw:
- 2.9% improvement in 10-km run time (controls: -0.1%)
- 11% improvement in submaximal cycling efficiency (controls: 1.4%)
- 8.5% drop in fat mass (controls: 2.6%)
A 2021 home-based replication by Rønnestad confirmed the effect with 55 cyclists. The sleep-low group gained 5.5% in FTP over three weeks versus 1.3% in controls. One caveat: low-intensity session power dropped 3% in week one as athletes adapted to training depleted. Hard session intensity was preserved throughout.
The Rønnestad data tells you how to dose it. The adjustment cost in week one is real but temporary.
The Chronic Low-Carb Trap: Why Keto Is Not the Same as Train-Low
This is where most consumer content fails athletes completely.
Some athletes read “train low” and decide to go ketogenic. Train hard, run low on carbs all the time, then eat a mountain of pasta the night before race day. The reasoning sounds logical. It’s wrong.
Burke’s 2020 PLOS ONE study of 26 elite racewalkers is definitive. The athletes on a chronic low-carb, high-fat diet improved fat oxidation dramatically — from 0.59 to 1.26 grams per minute. But their 10-km race performance dropped 2.3% and their oxygen cost at race pace rose 6.2%. The high-carb group improved performance 4.8%.
More fat burning, slower racing. The body became so adapted to fat that it couldn’t switch back to carbohydrate efficiently.
The 2021 Journal of Physiology follow-up closed the door completely. After keto adaptation, athletes could only restore carbohydrate oxidation to 61% of normal values at race pace and 78% at lower intensities — even after eating high-carb for acute restoration. The machinery that burns carbohydrate fast gets dismantled. You can’t rebuild it overnight.
Chronic keto in training costs you the ability to burn carbs at race intensity. That’s the one metabolic skill you need most on race day.
The risk doesn’t stop there. Pushing too many sessions per week into train-low territory compounds the fuel deficit and raises the risk of relative energy deficiency and hormonal disruption — a concern that’s especially serious for female athletes and those already training at high volume.
Take James, 42, Training for His First Sub-10 Ironman
James ran 10:24 at his last race. Nutrition was solid on the bike but he limped through the run — felt flat from kilometre 12 onward. He was already training 12 hours per week. More volume wasn’t an option.
Looking at his training week, two sessions stood out as ideal train-low candidates: a 60-minute easy run on Tuesday morning and a 50-minute easy spin on Thursday morning. Both followed hard evening sessions the night before. Both were low enough intensity that glycogen depletion wouldn’t compromise quality.
He kept Tuesday dinner carb-free after his Monday interval session. Same for Wednesday evening after a tempo ride. Over eight weeks, his submaximal efficiency improved noticeably — he was holding the same pace at a lower heart rate in those easy sessions. On race day, fully fueled from Friday, he ran the marathon leg 11 minutes faster. Final time: 10:13.
The gain came from two sessions per week. Not a diet overhaul.
Which Sessions Are Train-Low Candidates
The decision rule is simple: easy intensity plus short duration plus following a depleting session. That’s it.
| Session Type | Intensity | Duration | Train-Low? | Why |
|---|---|---|---|---|
| Easy recovery run | Z1-Z2 | Under 60 min | Yes | Low demand; strong adaptation signal |
| Base aerobic ride or run | Z2 | 60–90 min | Yes, with care | Protein before; no carbs during |
| Long endurance session | Z2 | Over 90 min | No | Bonk risk; immune window opens |
| Threshold or tempo intervals | Z3-Z4 | Any | No | Quality collapses; train high |
| VO2max or race-pace intervals | Z4-Z5 | Any | No | Carb-dependent; train high |
| Second session same day (2-a-day) | Z2 | Under 60 min | Yes | Designed for low glycogen |
| Key brick workout (triathlon) | Z2-Z4 mixed | Over 90 min | No | Full fueling throughout |
Anything touching threshold intensity or running longer than 90 minutes requires full fueling. Non-negotiable.
Easy sessions never go to waste as train-low candidates, but your Zone 2 training base still needs to be genuine Zone 2 — not so depleted that quality drops.
The Immune Risk You’re Underestimating
Train low too often and your immune system pays. Hard sessions already open a 3 to 72-hour window of immune suppression after every bout. Carbohydrate during exercise (a 6% solution at 1 litre per hour) attenuates this by blunting cortisol and IL-6 elevation. Train-low sessions skip this protection.
Gleeson et al. found that 33.3% of runners reported an upper respiratory tract infection after a 56-km race, versus 15.3% in non-running controls. High weekly training volume above 65 km was an independent risk factor.
Do two or three train-low sessions per week. Not six.
Impey et al. recommend 30 to 50% of total weekly sessions as the ceiling for reduced carbohydrate availability. For a six-session week, that’s two or three sessions maximum. The rest train high.
What a Practical Week Looks Like
The Prieto-Bellver 2025 Physiological Reports trial found something often missed: it’s not just the restriction phase that matters. A group that did four weeks of reduced carbohydrate followed by four weeks of reintroduction was the only group to improve running economy at lactate threshold (-27 ml per kg per km, p = 0.026). The chronic low-carb group and the chronic high-carb group both showed no improvement.
Translation: you restrict to build the adaptation, then reload to express it. The cycling matters.
A practical weekly structure:
- Monday: Rest or easy Z1 walk (optional train-low)
- Tuesday evening: Hard session — threshold intervals, fully fueled
- Wednesday morning: Easy 60-min run, train-low (no carbs after Tuesday dinner)
- Wednesday evening: Easy ride or swim, normal fueling
- Thursday: Zone 2 ride or run, fully fueled
- Friday: Short easy session, optional train-low
- Saturday: Long ride or long run, fully fueled with 60–90 g carbohydrate per hour
- Sunday: Recovery, full fueling for next week
This is the rhythm. Hard sessions eat carbs. Easy sessions after hard sessions don’t.
For race day, your carbohydrate loading targets should be locked in before race week starts — not improvised on the morning.
How AthleteOS Flags the Decision
The hardest part of carbohydrate periodization isn’t the science. It’s the weekly decision: which session is a train-low candidate and which one isn’t?
AthleteOS reads your weekly training plan and automatically flags sessions that sit in the candidate window — easy-intensity sessions under 90 minutes that follow a glycogen-depleting hard session the previous evening. It separately flags your key sessions (threshold intervals, race-pace bricks, long rides over 90 minutes) as full-fueling non-negotiables.
If you have back-to-back hard days, the system marks the intermediate easy session as a poor train-low candidate regardless of its intensity, because your recovery needs take priority. The flag shows up in your workout calendar before you pack your bag, not after you’ve already decided.
Sign up for AthleteOS to see which sessions in your current week are candidates.
The Bottom Line
Carbohydrate periodization works when the right sessions go low and the wrong ones never do.
The molecular signal is real — 73 to 78% of studies confirm it. The performance translation is selective — only 37% of performance studies show gains, and the studies that work are the ones where hard sessions stayed fully fueled.
Chronic keto is not periodization. Burke’s data ends that argument: CHO oxidation drops to 61% of normal after keto adaptation and doesn’t fully recover overnight. You can’t eat your way out of that suppression on race morning.
Train low on your easy sessions. Race high on everything else. Two or three sessions per week. Eat enough protein to protect lean mass on the low days. Reload with 8 to 10 grams of carbohydrate per kilogram before key sessions.
Two sessions per week. That’s all it takes.