../

Energy systems

How muscles remake ATP: the phosphagen, glycolytic and oxidative systems, how much each supplies at different effort durations, which fuel they burn, what lactate does, how to train each system, and how to keep energy up day to day. Running sessions are in endurance, lifting in strength training, and daily eating in nutrition and hydration. For work, power and energy as physics, see physics fundamentals.

ATP: the energy currency

Muscle contraction, calcium pumping and ion pumps all run on ATP splitting to ADP and phosphate. The stored ATP is tiny, so it has to be remade as fast as it is used.

FactValueSource
ATP stored in muscle≈ 5 mmol/kg wet muscleHargreaves & Spriet 2020
ATP use, all-out sprint (≈ 900 W)3.7 mmol/kg/s: stored ATP alone would last under 2 sHargreaves & Spriet 2020
ATP use, ≈ 75% VO₂max (≈ 200 W)0.4 mmol/kg/s: stored ATP would last ≈ 15 sHargreaves & Spriet 2020
ATP level in hard exercisefairly well defended: ≈ 20% lower in very intense work, 30% lower after a 30 s sprintHargreaves & Spriet 2020; Bogdanis et al. 1995

Ways to remake ATP

PathwayReaction (simplified)ATP gained
phosphagenPCr + ADP → ATP + creatine1 per PCr
glycolysis to lactateglycogen (1 glucose unit) → 2 lactate3
glucose oxidationglucose + 6 O₂ → 6 CO₂ + 6 H₂O≈ 36 by the traditional count; lower with modern estimates
fat oxidationpalmitate + 23 O₂ → 16 CO₂ + 16 H₂O≈ 130 by the traditional count; also lower

Yields from Hargreaves & Spriet 2020 (Box 1). The ratios between fuels hold whichever count you use.

The three systems

All three run all the time. What changes with intensity and duration is how much each one supplies. Fast systems have high rate (power) but small capacity (total ATP); slow systems are the reverse.

Phosphagen (ATP–PCr)Glycolytic ("anaerobic", fast glycolysis)Oxidative (aerobic)
wherecytosolcytosolmitochondria
fuelphosphocreatine (PCr)muscle glycogen, blood glucosecarbohydrate, fat (a little protein)
needs O₂?nonoyes
rate (power)highesthighlowest; carbohydrate faster than fat
capacitysmallest: PCr ≈ 75 mmol/kg dry muscle, mostly used in 10–15 s all-out≈ 3× PCr: ≈ 225 mmol/kg dry muscle over 30–90 svery large: glycogen, then fat
main rolefirst ≈ 10 s of maximal work; jumps, throws, lifts, sprints10 s to ≈ 1–2 min of hard workdominant beyond ≈ 75–80 s; everything submaximal
by-productscreatine, phosphate (Pi)lactate, H⁺CO₂, water
what limits itPCr runs outrising acidity, not lack of glycogenO₂ delivery and mitochondria; glycogen in long events
recoveryPCr half back in ≈ 20–60 s; near full in several minutesas PCr recovers and acidity clearsglycogen: ≈ 24 h to refill with enough carbohydrate

Sources: capacities, limits and timing from Hargreaves & Spriet 2020; crossover time from Gastin & Suppiah 2026; glycogen refill from Murray & Rosenbloom 2018; PCr recovery below.

PCr resynthesis

AfterPCr recoveredSource
exhaustive dynamic exercisebiphasic: fast half-time 21–22 s, slow half-time over 170 sHarris et al. 1976
a 30 s all-out cycle sprint65% at 1.5 min, 85.5% at 6 min; half-time ≈ 57 s; power recovers in step with PCrBogdanis et al. 1995
with blood flow blocked (cuff)none; resynthesis needs oxygenHarris et al. 1976

So a 6–10 s sprint needs 1.5–3 min rest to repeat near full power; after a 30 s effort, even 6 min is not quite enough. PCr is remade aerobically, so a better aerobic base speeds recovery between sprints and sets.

Who does what in a short sprint

EffortEnergy splitSource
first 6 s cycle sprintPCr and anaerobic glycolysis about equal; PCr fell 57%Gaitanos et al. 1993
30 s Wingate test18.6% aerobic, 31.1% phosphagen, 50.3% glycolyticBeneke et al. 2002
last 5 s of a 30 s sprint≈ 50% aerobic; VO₂ reaches 70–100% of VO₂maxHargreaves & Spriet 2020

The oxidative system runs in the mitochondria: the Krebs cycle in the matrix, the electron transport chain and ATP synthase on the folded inner membrane (cristae). Endurance training adds mitochondria.

Cut-away diagram of a mitochondrion showing the outer membrane, the folded inner membrane (cristae), the matrix, ATP synthase particles, ribosomes and DNA Image: Mariana Ruiz Villarreal (LadyofHats), public domain, via Wikimedia Commons (opens in a new tab).

Watch for how the three systems overlap rather than switch on in turn, and which fuel each uses.

Overview of 3 main energy systems used for exercise (opens in a new tab) (Jayson Gifford Exercise Physiology, YouTube)

Contribution by effort duration

For a single all-out effort, the aerobic share rises fast with duration. Gastin & Suppiah's 2026 systematic review (102 studies, 311 data points) put the 50:50 point at 78.6 s (≈ 75–80 s). Its estimates differ from Gastin's 2001 review by 0–3 percentage points.

share of energy supplied, % (approximate) 0 50 100 10 s 30 s 1 min 2 min 4 min 10 min duration of an all-out effort (log scale) phosphagen (ATP–PCr) glycolytic (“anaerobic”) oxidative (aerobic) aerobic = anaerobic at ≈ 75–80 s
Approximate energy-system shares in one all-out effort (aerobic line from Gastin & Suppiah 2026; the PCr/glycolytic split is a rough estimate)
All-out durationAerobicAnaerobic
10 s9%91%
20 s18%82%
30 s25%75%
60 s42%58%
75 s49%51%
2 min62%38%
4 min78%22%
10 min89%11%
15 min94%6%

Model estimates from Gastin & Suppiah 2026 (Table 5). The 95% prediction error is ± 12–14 points, so treat these as a guide, not a measurement. Running and cycling gave the same curve, and so did trained and untrained groups.

By running event

EventTypical timeAerobic shareSource, population
100 m10–12 s9–25% (depends on method; 21% men, 25% women by O₂ deficit)Duffield et al. 2004, trained track athletes
200 m20–25 s21–33% (Duffield); 29% (Spencer & Gastin)Duffield et al. 2004; Spencer & Gastin 2001
400 m45–60 s43%Spencer & Gastin 2001, highly trained
800 m≈ 2 min66%Spencer & Gastin 2001
1500 m≈ 4 min84% (treadmill); 77% men, 86% women (track time trials)Spencer & Gastin 2001; Duffield et al. 2005
3000 m8–10 min86% men, 94% womenDuffield et al. 2005
5 k13–30 min≥ 94%15 min all-out in Gastin & Suppiah 2026
10 k to marathon30 min to 5 h≈ 98% at 30 min, ≈ 99% at 1 h and beyondextrapolated from the Gastin & Suppiah 2026 model, fitted to efforts up to 15 min

Lab and field methods disagree by several points for sprints. Even the 800 m is mostly aerobic, so middle-distance runners need a big aerobic base as well as speed.

Fuel use by intensity

IntensityMain fuelSource
rest, easy (≤ 45% VO₂max, trained)mostly fatBrooks & Mercier 1994
FATmax, general adultspeak fat burning at 48% VO₂max (≈ 62% HRmax); men 45%, women 52%Venables et al. 2005, 300 adults
FATmax, trained cyclists64% VO₂max (74% HRmax); within 10% of peak from 55% to 72% VO₂maxAchten et al. 2002
hard (≈ 75% VO₂max)carbohydrate predominantBrooks & Mercier 1994
≥ 89% VO₂max (≈ 92% HRmax)fat contribution negligible; nearly all carbohydrateAchten et al. 2002
sprints and intervalsPCr and glycogenHargreaves & Spriet 2020

Crossover concept (Brooks & Mercier 1994): the crossover point is the intensity where carbohydrate takes over from fat as the main fuel. Harder exercise pushes toward carbohydrate. Endurance training pushes the other way: more mitochondria and less adrenaline at a given load mean more fat at the same pace.

PointDetail
peak fat oxidationon average 7.8 mg/kg fat-free mass/min; higher in women than men (Venables et al. 2005)
person to personsex, VO₂max and activity explained only 12% of the variation in peak fat oxidation (Venables et al. 2005)
burning fat vs losing fatweight change depends on energy balance over days, not on the fuel mix in one session
race pacein most Olympic events, endurance races included, carbohydrate is the main fuel (Hargreaves & Spriet 2020)

Fuel stores

StoreTypical amountEnergySource
muscle glycogen≈ 500 g (range 300–700 g)≈ 2 000 kcal (4 kcal/g)Murray & Rosenbloom 2018
liver glycogen≈ 80 g (range 0–160 g); falls overnight≈ 320 kcalMurray & Rosenbloom 2018
blood glucose≈ 4 g≈ 16 kcalMurray & Rosenbloom 2018
whole-body glycogen≈ 600 g; varies with body mass, diet, training≈ 2 400 kcalMurray & Rosenbloom 2018
muscle PCr≈ 75 mmol/kg dry muscle≈ 10–15 s all-outHargreaves & Spriet 2020
body fatkilograms, even in lean athletestens of thousands of kcal: effectively unlimited for exercise
Glycogen factValueSource
breakdown rate, all-outup to 40 mmol glucose/kg wet muscle/minMurray & Rosenbloom 2018
breakdown rate, easy1–2 mmol glucose/kg wet muscle/minMurray & Rosenbloom 2018
supercompensation (loading)8–10 g carbohydrate/kg/day for 24–72 h with rest or very light trainingMurray & Rosenbloom 2018
loading before events over 90 min10–12 g/kg/day for 36–48 hThomas et al. 2016
what limits a 30–90 s effortacidity, not glycogen running outHargreaves & Spriet 2020
refillcomplete in ≈ 24 h with ≈ 10 g/kg carbohydrate; 1.0–1.2 g/kg/h in the first hours after exercise speeds itMurray & Rosenbloom 2018

"Hitting the wall" in a marathon is running low on muscle and liver glycogen. Race fueling is in endurance.

Lactate and thresholds

MythWhat the evidence saysSource
lactate is waste made only without oxygenmade all the time, even at rest with enough O₂; a major fuel (heart, slow fibers), the main raw material for new glucose, and a signalBrooks 2018
lactic acid causes the burn and acidosisthe H⁺ comes from ATP splitting when glycolysis and PCr supply the ATP; making lactate actually consumes H⁺ and slows acidosis (Robergs' view; some still debate the accounting)Robergs et al. 2004
lactate causes fatigueacidosis has little direct effect at body temperature; phosphate from PCr breakdown is a bigger causeWesterblad et al. 2002

Blood lactate is still a useful marker of how hard the muscle is working.

TermDefinitionTypical value
LT1 (aerobic threshold, VT1)first rise in blood lactate above baseline1.0–2.0 mmol/L in trained athletes (Seiler-Viken et al. 2025)
LT2 (anaerobic threshold, VT2, respiratory compensation)second, steeper rise≈ 2.5–4.0 mmol/L, varies with the athlete (Seiler-Viken et al. 2025)
MLSShighest workload where lactate stays steady; in a constant 30 min test, a rise of ≤ 1 mmol/L from minute 10 to 30usually sustainable 30–60 min; group means 4.3–5.1 mmol/L depending on the method (Beneke 2003; Beneke et al. 2011)
OBLA, "4 mmol"fixed 4 mmol/L pointa population average, not your threshold
  • LT2 and MLSS mark the top of steady-state exercise. Field estimate: the pace you could race for about 30–60 min, since MLSS is usually sustainable that long (Beneke et al. 2011).
  • LT1 is the top of easy running; the talk-test and zone definitions are in endurance.
  • Beneke et al. 2011 found no evidence that training exactly at threshold works better than other intensities.

VO₂max, economy and EPOC

Endurance performance rests on three factors (Joyner & Coyle 2008):

FactorMeaningWhat improves it
VO₂maxhighest rate of O₂ use (mL/kg/min); the size of the aerobic engineintervals at 90–100% VO₂max, plus volume
fractional use (threshold)% of VO₂max you can hold, set by LT2 / MLSSthreshold work, easy volume
economy (efficiency)O₂ cost at a given speed or poweryears of running, strength and plyometrics, shoes
race speed≈VO2max×fraction heldO2 cost per km\text{race speed} \approx \frac{\text{VO}_2\text{max} \times \text{fraction held}}{\text{O}_2\text{ cost per km}}

Two runners with the same VO₂max can race very differently if one holds a higher fraction or runs more economically. Economy is the least understood of the three (Joyner & Coyle 2008).

EPOC ("afterburn")

PointDetailSource
what it pays forrefilling O₂ stores, remaking ATP and PCr, clearing lactate, higher temperature, breathing and circulationBørsheim & Bahr 2003
shapea fast part (minutes) and a slow part (up to hours)Børsheim & Bahr 2003
size vs intensityrises steeply (roughly exponentially) with intensity; roughly linearly with duration above 50–60% VO₂maxLaForgia et al. 2006
long EPOC (3–24 h)only after ≥ 50 min at ≥ 70% VO₂max, or ≥ 6 min total at ≥ 105% VO₂maxLaForgia et al. 2006
how big6–15% of the net O₂ cost of the exercise itselfLaForgia et al. 2006
fat lossminor; the energy spent during exercise is what countsLaForgia et al. 2006

Training each system

The classic NSCA work:rest ratios, by the system you want to load:

Target systemIntensity (% of max power)Work boutWork:restExample
phosphagen (alactic power)90–100%5–10 s1:12 to 1:208 s sprint, 1.5–2.5 min walk
fast glycolysis75–90%15–30 s1:3 to 1:530 s hard, 1.5–2.5 min easy
glycolysis + oxidative30–75%1–3 min1:3 to 1:42 min hard, 6 min easy
oxidative20–35%over 3 min1:1 to 1:34 min at 5 k pace, 2–4 min jog

Source: NSCA Essentials of Strength Training and Conditioning, bioenergetics chapter. Percentages are of maximal power output, not of VO₂max.

GoalSession typesAdaptations
alactic power, speedshort sprints, jumps, heavy singles with full restfaster PCr use; neural drive
repeat-sprint ability6 s sprints with 20–30 s restfaster PCr recovery (needs aerobic fitness)
glycolytic capacity, "lactate tolerance"20–60 s efforts, 1:3–1:5 restbuffering, higher peak lactate; very fatiguing, so use sparingly
VO₂max3–5 min at 90–100% VO₂max, 1:1 reststroke volume, VO₂max
threshold10–20 min blocks at LT2 pace, or 30–40 min continuoushigher LT2 / MLSS
aerobic baselong, easy (below LT1), high volumemitochondria, capillaries, fat use, economy

Elite endurance athletes do about 80% of sessions at low intensity (≈ 2 mmol/L lactate or below) and 20% hard (Seiler 2010). Sessions and zones are in endurance, blocks and deloads in training plans.

Keeping energy up day to day

Energy availability and REDs

EA=energy intake−exercise energy expenditurefat-free mass (kg)kcal/kg FFM/day\text{EA} = \frac{\text{energy intake} - \text{exercise energy expenditure}}{\text{fat-free mass (kg)}} \quad \text{kcal/kg FFM/day}
PointDetailSource
healthy reference≈ 45 kcal/kg FFM/day is linked with normal body functionIOC REDs consensus (Mountjoy et al. 2023)
low energy availabilitymany body systems are disturbed below ≈ 30 kcal/kg FFM/day (mostly data from women); this is not a universal cut-offMountjoy et al. 2023
REDsRelative Energy Deficiency in Sport: too little energy for training, hitting hormones, bone, immunity, mood and performance, in women and menMountjoy et al. 2023
low carbohydrate availabilityemerging data: may do harm of its own, on top of low energyMountjoy et al. 2023
warning signsmissed or stopped periods, low libido, bone stress injuries, frequent illness, stalled performance, low moodMountjoy et al. 2023

What to eat, how much carbohydrate and protein, and when, is in nutrition and hydration.

Sleep

PointDetailSource
adults7 h or more per night, regularlyAASM/SRS (Watson et al. 2015)
athleteselite athletes often sleep under 7 h; a fixed 7–9 h rule may not fit everyone, so work from your own sleep needWalsh et al. 2021
why it mattersa night without sleep cuts performance; habitual sleep under 7 h raises the risk of respiratory infectionWalsh et al. 2021

Caffeine

PointDetailSource
effective dose3–6 mg/kg about 60 min before; may work from 2 mg/kgISSN (Guest et al. 2021)
too much≈ 9 mg/kg brings frequent side effects and no extra benefitISSN (Guest et al. 2021)
strongest benefitaerobic endurance; also strength, sprinting and alertnessISSN (Guest et al. 2021)
safe intake, healthy adultsup to 400 mg/day, single doses up to 200 mg; pregnancy up to 200 mg/dayEFSA 2015
sleep400 mg taken 6 h before bed still cut total sleep timeDrake et al. 2013
gels, gumscaffeinated gum acts faster than capsulesISSN (Guest et al. 2021)

A 70 kg adult at 3 mg/kg takes 210 mg, about the EFSA single-dose level. Higher doses are for races, not daily training.

Recipes

Alactic power session

For speed and power without much fatigue: short, maximal, fully recovered. Keep the phosphagen system as the main supplier by stopping each rep before 10 s.

Warm-up    15 min easy + drills + 3 build-ups (see running warm-up sheet)
Main       2 sets × 5 × 8 s hill sprint (or 6 s flat, or 6 s bike sprint)
Rest       walk back, 2 min between reps (≈ 1:15), 4–5 min between sets
Stop when  a rep is clearly slower or you lose form
Cool-down  10 min easy
Total hard work: 80 s

Glycolytic interval session

For 400–1500 m speed or finishing kicks. Very taxing: once a week at most, never in the week before a key race.

Warm-up    15–20 min easy + drills + 4 strides
Main       6–8 × 30 s at 85–90% effort (≈ 400–800 m race pace)
Rest       2 min walk or very easy jog (≈ 1:4)
Stop when  pace drops over 5% on two reps in a row
Cool-down  15 min easy (it helps clear lactate faster)

Aerobic base week

For building the oxidative system: mostly easy, one quality session, one long run. Easy means you could talk in full sentences.

Mon  rest or 30 min easy cycle
Tue  45 min easy + 6 × 20 s strides
Wed  lift (see strength training)
Thu  50 min: 3 × 8 min at threshold, 2 min jog
Fri  rest
Sat  40 min easy + lift
Sun  long run 75–90 min easy
≈ 80% of running time below LT1, ≈ 20% at or above threshold

Estimate lactate threshold from a 30 min time trial

In McGehee et al. 2005, the 30 min time trial estimated running speed at the 4 mmol/L threshold with a standard error of 0.21 m/s, and heart rate with 8 bpm.

  1. Warm up 15 min with 3–4 strides. Flat course or track, calm weather, rested.
  2. Run 30 min as far as you can at an even effort, alone.
  3. Threshold pace ≈ your average pace for the 30 min.
  4. Threshold heart rate ≈ your average heart rate for the last 20 min (common field practice; heart rate rises in the first 10 min).
  5. Retest every 6–8 weeks, on the same course.
Distance in 30 min: 7.0 km
Threshold pace:     30 min ÷ 7.0 km = 4:17 /km
Average HR, min 10–30: 168 bpm  → threshold HR ≈ 168
Threshold session:  3 × 10 min at 4:15–4:20 /km, 2 min jog

Fuel planning by session

A quick table for what the session will burn and what to eat around it. Carbohydrate amounts during exercise follow Jeukendrup 2014 and Thomas et al. 2016.

SessionMain fuelBeforeDuring
easy run under 60 minfat and carbohydrate mixnormal meals; fasted is fine if it suits youwater if hot
sprints, liftingPCr, glycogena normal meal 2–4 h beforewater
intervals, threshold, 45–75 minmostly carbohydratecarbohydrate meal 2–4 h beforewater; a sports drink or carbohydrate mouth rinse can help
long run 1–2.5 hcarbohydrate, with fat rising over time1–4 g/kg carbohydrate in the 1–4 h before30–60 g carbohydrate/h
race or long run over 2.5 hcarbohydrateload 36–48 h before (10–12 g/kg/day)up to 90 g/h of glucose plus fructose

References