The focus of positive selection studies in humans needs to move from candidate locus discovery to pinpointing underlying causal variants and further investigation of their biological significance. We performed a meta-analysis of published selection screens and extended these with an analysis of the 1000 Genomes Project Phase 3 SNP dataset using our newly developed method of Fine-Mapping of Adaptive Variation (FineMAV). The FineMAV score combines population differentiation, derived allele frequency and a measure of molecular functionality to produce a refined list of candidate variants for functional follow-up. We calibrated and tested FineMAV using eight ‘gold standard’ examples of experimentally-validated causal variants underlying adaptations, and were able to pick out the known functional allele in all instances. We used this approach to identify the best candidate variants driving positive selection in Africans, Europeans, East and South Asians, and report many novel examples including rs6048066 in TGM3 associated with curly hair, and rs7547313 in SPTA1 associated with erythrocyte shape and possibly malaria resistance in Africans. We also picked up rs201075024 in PRSS53 associated with hair shape in South Asians. FineMAV now offers a better way to identify specific variants for functional follow-up and paves the way for identification of causative alleles driving phenotypic differences among human populations.