Mitochondrial respiratory complexes form superassembled structures called supercomplexes. COX7A2L is a supercomplex-specific assembly factor in mammals, although its implication for supercomplex formation and cellular metabolism remains controversial. Here we identify a role for COX7A2L for mitochondrial supercomplex formation in humans. By using human cis-expression quantitative trait loci data, we highlight genetic variants in the COX7A2L gene that affect its skeletal muscle expression specifically. The most significant cis-expression quantitative trait locus is a 10-bp insertion in the COX7A2L 3' untranslated region that increases messenger RNA stability and expression. Human myotubes harboring this insertion have more supercomplexes and increased respiration. Notably, increased COX7A2L expression in the muscle is associated with lower body fat and improved cardiorespiratory fitness in humans. Accordingly, specific reconstitution of Cox7a2l expression in C57BL/6J mice leads to higher maximal oxygen consumption, increased lean mass and increased energy expenditure. Furthermore, Cox7a2l expression in mice is induced specifically in the muscle upon exercise. These findings elucidate the genetic basis of mitochondrial supercomplex formation and function in humans and show that COX7A2L plays an important role in cardiorespiratory fitness, which could have broad therapeutic implications in reducing cardiovascular mortality.