Abstract Individual levels of physical activity, and especially of voluntary physical exercise, highly contribute to the susceptibility for developing metabolic, cardiovascular diseases, and potentially to psychiatric disorders. Here, we applied a cross-species approach to explore a candidate genetic region for voluntary exercise levels. First, a panel of mouse chromosome substitution strains was used to map a genomic region on mouse chromosome 2 that contributes to voluntary wheel running levels – a behavioral readout considered a model of voluntary exercise in humans. Subsequently, we tested the syntenic region (HSA20: 51,212,545–55,212,986) in a human sample (Saint Thomas Twin Register; n=3038) and found a significant association between voluntary exercise levels (categorized into excessive and non-excessive exercise) and an intergenic SNP rs459465 (adjusted P-value of 0.001). Taking under consideration the methodological challenges embedded in this translational approach in the research of complex phenotypes, we wanted to further test the validity of this finding. Therefore, we repeated the analysis in an independent human population (ALSPAC data set; n=2557). We found a significant association of excessive exercise with two SNPs in the same genomic region (rs6022999, adjusted P-value of P=0.011 and rs6092090, adjusted P-value of 0.012). We explored the locus for possible candidate genes by means of literature search and bioinformatics analysis of gene function and of trans-regulatory elements. We propose three potential human candidate genes for voluntary physical exercise levels (MC3R, CYP24A1, and GRM8). To conclude, the identified genetic variance in the human locus 20q13.2 may affect voluntary exercise levels.