Ideal Weight vs BMI vs FFMI for a 178cm Male: Three Failure Modes

"Ideal weight" formulas predate body-composition measurement. Devine, Robinson, Miller, and Hamwi were derived in the 1960s-80s from clinical populations to set drug-dosing or insurance-pricing benchmarks. Applied to trained lifters, they produce wrong answers in a predictable direction. This article runs the three engines on a 178 cm male and shows where each one fails.

Engine outputs

The 178 cm male's current 80 kg state through the BMI Calculator and the FFMI Calculator (16% body fat). The four Ideal Weight Calculator formulas are derived by hand below, since that engine takes only height and sex (a 178 cm male is its built-in demo case):

# bmi-calculator (computed live from /engines/bmi-calculator.js)
Engine input
  weight_kg             = 80
  height_cm             = 178

Engine output
  bmi                   = 25.24933720489837
  rangeLabel            = Above reference range

# ffmi-calculator (computed live from /engines/ffmi-calculator.js)
Engine input
  weight_kg             = 80
  height_cm             = 178
  body_fat_pct          = 16

Engine output
  ffmi                  = 21.20944325211463
  adjustedFfmi          = 21.33144325211463
  fatFreeMassKg         = 67.2
  interpretation        = Above average — consistent training

The Ideal Weight Calculator returns Devine 73.18 kg, Robinson 71.15 kg, Miller 70.41 kg, Hamwi 75.21 kg. Working Devine by hand: 50 + 2.3 × (inches over 60). For 178 cm = 70.08 inches, that is 50 + 2.3 × 10.08 = 73.18 kg. The other formulas use similar height-based linear schemes with different coefficients — a 4.8 kg spread.^[1]

At the current 80 kg the BMI calculator returns 25.25 ("Above reference range"); rerun at the Devine "ideal" of 73 kg it returns 23.04 and at 70 kg it returns 22.09, both inside the reference range. FFMI at 80 kg / 16% body fat is 21.21 ("Above average — consistent training"); at 73.2 kg with the same body fat it drops to 19.41 ("Average muscle mass") and at 70 kg to 18.56.

Reading the disagreement

The Ideal Weight engine says "73.2 kg." The BMI calculator confirms 73.2 kg sits inside the reference band. The FFMI calculator says cutting from 80 to 73.2 kg drops FFMI from 21.2 to 19.4, moving the lifter from "Above average" into "Average."

A trained lifter cutting from 80 kg to the Devine 73.2 kg target loses 6.8 kg. At constant 16% body fat, about 5.7 kg of that is lean tissue (0.84 × 6.8). That is what drives FFMI from 21.2 down to 19.4 — the "ideal weight" target is asking the lifter to give back roughly two FFMI points of training adaptation.^[3]

Where Devine and friends actually fail

Body composition invisibility

Devine and the related formulas were derived to estimate the lean body mass of a "normal" patient for drug-dosing purposes. They assume an implicit body-fat percentage near 20% — the population mean of the cohorts they were fitted on. For a lifter who carries more lean mass than that average, the "ideal weight" target lands below the lifter's actual healthy weight: here the 73.2 kg Devine figure sits 6.8 kg under the lifter's current 80 kg, even though the lifter is at 16% body fat and would lose mostly muscle reaching it.^[1]

Single-variable input

All four formulas use height as the sole input. Sex is a binary modifier (lower constant for women), but bodyweight, body composition, and training status do not enter the math. Two 178 cm men of vastly different builds get the same "ideal weight."

Historical context

Devine published his formula in 1974 as a quick-reference for aminoglycoside antibiotic dosing. He never claimed it predicted health outcomes. Its use as a "target body weight" is a 1980s-1990s retconning that the original literature does not support.^[1]

Where BMI fails

BMI's failures are well-documented and largely repeat the Devine failure modes: invisible to body composition, single-input (weight and height only), and validated for population-level mortality rather than individual classification. The BMI engine returns "Above reference range" for the 80 kg case because the population-derived band ends at 24.9.^[2]

For lifters with FFMI above 20, BMI fires a false-positive overweight label at any reasonable trained-population weight. The published clinical recommendation is the same one used in the BMI-vs-FFMI piece: use BMI for screening, not diagnosis, and run a body-composition check whenever the BMI flag fires for an active person.

How FFMI quietly replaces both

FFMI does what Devine and BMI were trying to do but with the missing piece (lean mass) baked into the math. It normalises for height, surfaces body composition directly, and produces a number that maps onto a meaningful endpoint (the natural-cap distribution from the Kouri 1995 paper).^[3]

For a trained lifter, the FFMI percentile against the natural-trained-male distribution is a more meaningful "where am I?" reading than the Devine-implied target weight. The 178 cm / 80 kg / 16% body fat lifter at FFMI 21.2 sits in a healthier band than the same lifter would at the Devine target of 73.2 kg with the same body fat.

When the older tools are still right

1. Drug dosing. Devine and Robinson are still used clinically for aminoglycoside antibiotic dosing because the formula's implicit lean-mass estimate is good enough for that purpose. This is the original use case.
2. Population health screening. BMI's mortality association is strong enough for population epidemiology; the Devine/Hamwi formulas have no role here.
3. Untrained adult baseline. For sedentary adults outside the lifting population, the disagreement between Devine and BMI narrows because the implicit body-fat assumption (~20%) is closer to truth.

A decision frame for the trained lifter

1. Don't use Ideal Weight formulas for goal-setting. They under-predict for lifters with measurable training adaptation.
2. Treat BMI flags as a screening prompt, not a diagnosis. Run an FFMI or body-fat check first.
3. Use FFMI as the trained-population anchor. The natural-cap framing produces a meaningful target without the obsolete-formula penalty.
4. For body-fat goal weight specifically, work backward from a target body fat × current lean mass, not from a height-based formula.

Worked target-weight calculation

Concrete example for the 178 cm lifter targeting FFMI 23 at 14% body fat (a realistic advanced-intermediate target): lean mass at FFMI 23 = 23 × 1.78² = 72.9 kg. Bodyweight at 14% body fat with 72.9 kg lean = 72.9 / 0.86 = 84.8 kg. Devine says "ideal" is 73.2 kg. The realistic trained-lifter target is 84.8 kg. The Devine target is 11.6 kg below where the well-trained lifter wants to land.

Related reading

Natural Muscular Potential: FFMI develops the natural-cap framing this article leans on, Body Composition For Athletes sets out the trained-population context, and How To Build Muscle As A Beginner shows where the height-based "ideal weight" framing starts to break for someone adding lean mass.

FAQ

If I'm 80 kg at 178 cm and lifting, what should my weight goal be?

Set the goal based on body fat and FFMI, not Devine. At FFMI 21.2, the lifter is already above population average. A reasonable next-target FFMI is 23 (advancing toward the upper natural range) at the same body fat, which requires gaining 4–5 kg of lean mass rather than cutting weight.^[3]

Why does Hamwi return the highest "ideal weight"?

Hamwi (1964) was derived from a slightly more muscular insurance cohort than Devine (1974). The 4.8 kg spread across the four formulas comes from cohort differences and small math choices; none of the formulas was derived for lifters.^[1]

Is there a modern "ideal weight" formula that uses body fat?

No widely-used one. The contemporary replacement is FFMI-based target setting: pick a target FFMI in the realistic range (19–23 for recreational lifters, 22–25 for advanced), pick a target body fat, and compute the implied bodyweight from those two numbers.

What about waist-to-height ratio as an alternative to BMI?

Better than BMI for general adult mortality screening. For trained lifters, it adds little over body fat percentage because both numbers track fat mass with similar resolution. Use FFMI plus body fat for the trained population; reserve WHtR for general-population health screening.^[2]