Brown adipose tissue (BAT) is a promising therapeutic target against obesity. Therefore, research on the genetic architecture of BAT could be key for the development of successful therapies against this complex phenotype. Hypothesis-driven candidate gene association studies are useful for studying genetic determinants of complex traits, but they are dependent upon the previous knowledge to select candidate genes. Here, we predicted 107 novel-BAT candidate genes in silico using the uncoupling protein one (UCP1) as the hallmark of BAT activity. We first identified the top 1% of human genes predicted by the human gene connectome to be biologically closest to the UCP1, estimating 167 additional pathway genes (BAT connectome). We validated this prediction by showing that 60 genes already associated with BAT were included in the connectome and they were biologically closer to each other than expected by chance (p < 2.2 × 10−16). The rest of genes (107) are potential candidates for BAT, being also closer to known BAT genes and more expressed in BAT biopsies than expected by chance (p < 2.2 × 10−16; p = 4.39 × 10–02). The resulting new list of predicted human BAT genes should be useful for the discovery of novel BAT genes and metabolic pathways.