One of the major questions in ecology and evolutionary biology is how variation in the genome enables species to adapt to divergent environments. Here, we study footprints of thermal selection in candidate genes in six wild populations of the afrotropical butterfly Bicyclus anynana sampled along a c. 3000 km latitudinal cline. We sequenced coding regions of 31 selected genes with known functions in metabolism, pigment production, development and heat shock responses. These include genes for which we expect a priori a role in thermal adaptation and, thus, varying selection pressures along a latitudinal cline, and genes we do not expect to vary clinally and can be used as controls. We identified amino acid substitution polymorphisms in 13 genes and tested these for clinal variation by correlation analysis of allele frequencies with latitude. In addition, we used two FST-based outlier methods to identify loci with higher population differentiation than expected under neutral evolution, while accounting for potentially confounding effects of population structure and demographic history. Two metabolic enzymes of the glycolytic pathway, UGP and Treh, showed clinal variation. The same loci showed elevated population differentiation and were identified as significant outliers. We found no evidence of clines in the pigmentation genes, heat shock proteins and developmental genes. However, we identified outlier loci in more localized parts of the range in the pigmentation genes yellow and black. We discuss that the observed clinal variation and elevated population divergence in UGP and Treh may reflect adaptation to a geographic thermal gradient.