In recent years, there have been major developments of population genetics methods to estimate both rates of recombination and levels of natural selection. However, genomic variants subject to positive selection are likely to have arisen recently and, consequently, had less opportunity to be affected by recombination. Thus, the two processes have an intimately related impact on genetic variation, and inference of either may be vulnerable to confounding by the other. We illustrate here that even modest levels of positive selection can substantially reduce population-based recombination rate estimates. We also show that genome-wide scans to detect loci under recent selection in humans have tended to highlight loci in regions of low recombination, suggesting that confounding by recombination rate may have reduced the power of these studies. Motivated by these findings, we introduce a new genome-wide approach for detecting selection, based on the ratio of pedigree-based to population-based estimates of recombination rate. Simulations suggest that our “Ped/Pop” method, which is designed to capture completed sweeps, has good power to discriminate between neutral and adaptive evolution. Unusually for a multimarker method, our approach performs well in regions of high recombination and also has good power for many generations after the fixation of an advantageous variant. We apply the method to human HapMap and Perlegen data sets, finding confirmation of reported candidates as well as identifying new loci that may have undergone recent intense selection.