Does Running Really Cost 1 kcal per kg per km?

The number is exactly right inside the ACSM equation, roughly right in a laboratory, and its most famous property is a feature of the model before it is a fact about the body.

By Marc Ferrer · Founder, Engram Kinetics · Published September 2026 · Updated October 2026

This is educational content for certification candidates and coaches, not medical advice or individualized prescription. Equations are drawn from ACSM's published metabolic calculations; confirm coefficients against your current edition.

Where the number comes from

Start with the ACSM equation for level running, with speed S in m·min⁻¹:

VO₂ (mL·kg⁻¹·min⁻¹) = 3.5 + (0.2 × S)

Covering one kilometre takes 1000/S minutes. Multiply the rate by the duration, convert millilitres to litres, and apply an energy equivalent of 5 kcal per litre of oxygen:

GROSS = (3.5 + 0.2S) × (1000/S) / 1000 × 5 = 17.5/S + 1.0
NET   = (0.2S) × (1000/S) / 1000 × 5 = 1.000

In the net line the speed cancels. Not approximately: algebraically, leaving a constant. The famous rule of thumb and the ACSM running equation are the same statement written two ways.

What that gives you at real speeds

The table below starts at 134 m·min⁻¹, the speed above which ACSM presents the running equation as most accurate (5 mi·h⁻¹). Below that, the equation is outside the range its own source claims for it.

Values in kcal·kg⁻¹·km⁻¹, level running.

A worked example

A 70 kg runner covers 10 km at 12 km/h. Net cost is 70 × 10 × 1.000 = 700 kcal. Gross cost is 70 × 10 × 1.0875 = 761 kcal. The 61 kcal difference is simply resting metabolism over the 50 minutes the run took.

Why "independent of speed" needs care

Inside the ACSM equation, the net cost per kilometre cannot depend on speed. The equation is linear in S, so dividing the horizontal term by distance removes S entirely. The constancy is built in.

This matters because the equation is often cited as evidence that running cost per distance is speed-independent. It cannot be. A model that assumes a property cannot demonstrate it. The evidence has to come from measurement.

And it largely does, with qualifications. Track running measured with lactate correction gives a cost of 3.72 ± 0.24 J·kg⁻¹·m⁻¹, reported as independent of speed. That is 0.889 kcal·kg⁻¹·km⁻¹.

More recent work measured 56 recreational runners with a portable analyser, subtracting resting values and correcting the track trials for air resistance. It found a net cost of 4.20 ± 0.45 J·kg⁻¹·m⁻¹ on track and 4.35 ± 0.55 on a treadmill (1.004 and 1.040 kcal·kg⁻¹·km⁻¹). Those authors set out an expected band for a speed-independent cost of 3.5 to 4.5 J·kg⁻¹·m⁻¹, or 0.84 to 1.08 kcal·kg⁻¹·km⁻¹, which is probably the most honest single answer to the question in the title. They are conference proceedings and the authors describe the work as preliminary and ongoing, so weight them accordingly.

Two details from that study matter. Because the cost was computed net and corrected for air resistance, the rise they observed at the low and high ends of the speed range is not an artifact of resting metabolism; it is a genuine departure from flatness. And a 2022 study goes further still, finding that more precise laboratory measurement reveals an energy-optimal running speed rather than a flat curve.

So the honest position is: approximately constant across a mid-range of speeds, with measurable speed dependence at the edges and under finer measurement.

Comparing like with like

Cost-of-running figures are only comparable when you know what each one includes. The ACSM coefficient converts to 4.180 J·kg⁻¹·m⁻¹, which is 12% above the lactate-corrected track value and within half a percent of the more recent track measurement. Which figure looks like the outlier depends entirely on which you treat as the reference.

The ± on the 1993 figure is a standard deviation across 58 measurements, not a confidence interval.

Where the rule breaks

Gradient

The gradient term behaves the same way algebraically (speed cancels again) and leaves a clean correction:

extra net cost = 4.5 × G kcal·kg⁻¹·km⁻¹  (G as a decimal)

A 1% grade adds 0.045, a 5% grade adds 0.225, and a 10% grade adds 0.45; that last one is a 45% increase over level running. The equation does not handle downhill running, where mild descents cost less than level but steeper ones cost more because of braking.

The energy equivalent

The 5 kcal per litre figure depends on the respiratory exchange ratio. Across the range typical of running it moves from about 4.80 kcal·L⁻¹ at an RER of 0.80 to 5.05 at 1.00, which puts the net cost between 0.96 and 1.01 rather than at a fixed 1.000. Above the point where anaerobic energy contributes meaningfully, an oxygen-only calculation undercounts total expenditure regardless.

Air resistance

Treadmill running has no air resistance to overcome. Outdoors the penalty is small at jogging speeds and becomes material as speed rises, and a headwind changes the arithmetic considerably.

Individual variation

Running economy differs between people by more than the discrepancies discussed above. Expressing cost per kilogram of total body mass also distorts comparisons where body composition differs, since fat mass adds load without contributing to oxygen consumption. Treat 1.0 as a population estimate, not a personal measurement.

Why walking is different

Walking does not obey a constant cost per distance. Its cost curve is U-shaped, with a clear minimum at an intermediate speed and a sharp rise on either side. Reported optima differ by study and by whether the cost measured is net or gross, falling somewhere in the region of 1.06 to 1.30 m·s⁻¹. The spread itself reflects that these are different measurements, not one disputed number.

That difference in shape is the real contrast. Running cost per distance is flat over a broad middle range; walking cost per distance has a well-defined optimum. People do switch from walking to running somewhere around 2.0-2.2 m·s⁻¹, but that transition is not simply the point where one curve crosses the other: perceived effort and gait mechanics are part of it. The shapes explain why a single number works for one gait and not the other.

What to take away

• Net level running: 1.0 kcal·kg⁻¹·km⁻¹ is a sound working figure. Measured values sit in a band of roughly 0.84 to 1.08.
• Gross: add 17.5/S, which is 5 to 13% more depending on pace.
• Hills: add 4.5 × grade.
• Don't double-count resting metabolism when adding a run to your daily total.
• The exactness is the model's, not the body's. Inside the equation the answer is 1.000; in a laboratory it is approximately 1, and the approximation is the honest part.

Disclosure: Marc Ferrer is the founder of Engram Kinetics.

Independent exam-prep education, not affiliated with ACSM, NSCA or CSEP, and not medical advice. Confirm coefficients and conversions against your current edition before relying on them. Published by Engram Kinetics, engramkinetics.com.