Give 481 sedentary adults the exact same training programme for 20 weeks. Some will improve their VO2max by less than 5%. Others will improve by over 50%. The programme was identical. The difference is almost entirely genetic.
This is the central finding of the HERITAGE Family Study (Bouchard and colleagues, 1999), the most important investigation of genetically-determined training response heterogeneity ever conducted. 98 families, standardised cycling training 3 days per week at precisely controlled intensity. The range in VO2max response was extraordinary — from near-zero responders to 2.5× improvement relative to the mean.
---
Critically, when Bouchard separated co-twins in the data, he found that VO2max training responses within a twin pair were strikingly similar — even when the magnitude of average response across families varied enormously. This heritability of training responsiveness, as distinct from baseline fitness, is estimated at approximately 47% in the HERITAGE data — meaning nearly half the variance in how much you improve from identical training is determined by your genetic makeup before you even start.
This has a counterintuitive implication. High genetic potential does not mean high trainability. The correlation between baseline VO2max and the magnitude of VO2max improvement from training is actually weakly negative — meaning individuals who start with higher aerobic capacity tend to show smaller relative improvements from identical training. The high-responders in HERITAGE were often those with lower initial fitness.
---
Modern genome-wide association studies (GWAS) have attempted to identify the specific variants responsible for this training response heterogeneity. The results are humbling. Despite biobank sample sizes now exceeding 400,000 subjects (Klimentidis and colleagues 2020), polygenic scores for aerobic trainability and athletic performance achieve R² values of only 2–5% — meaning the current genetic data explains a tiny fraction of the variation that HERITAGE demonstrated exists.
The explanation is that athletic performance is highly polygenic — distributed across thousands of common variants each contributing very small effects. No single gene, no small panel of SNPs, and no consumer genomics test can meaningfully predict athletic ceiling. The DTC (direct-to-consumer) sports genomics industry claims more predictive validity than the peer-reviewed science supports.
---
What the HERITAGE data does usefully predict is that programme design should accommodate heterogeneity. Athletes who show minimal aerobic adaptations after 8–12 weeks of standard training are not necessarily undertrained — they may be low VO2max responders for whom the dominant training stimulus needs to shift toward factors where they have higher genetic responsiveness (muscular economy, lactate threshold, neuromuscular power).
The genetic architecture of gene-environment interaction (GxE) also explains why training responses appear to cluster in families — experienced coaches working across generations of athletes from the same family often informally notice that siblings respond similarly to training loads, and now we have the mechanistic science to understand why.
---
For athletes wanting to understand the interaction between their structural potential and current training progress, the genetic potential calculator at winsport.uk/tools/strength/genetic-potential-calculator uses anthropometric predictors to estimate structural athletic ceilings — providing a framework for understanding how much of the performance gap is physiological potential versus training and nutritional optimisation.
Does your programming account for individual response heterogeneity, or do all your athletes follow the same periodisation template?