Our understanding of the process of speciation has increased greatly since Darwin first proposed a central role for natural selection. Much of what we now know has come from research conducted over the past two decades. The picture emerging is that speciation is a process that results from the same forces responsible for most change within species: natural and sexual selection. Nonetheless, there are still many areas that require investigation.The ‘top down’ or phenotypic approach to studying speciation has found evidence for selection on ordinary phenotypic characters shown also to underlie premating and postmating isolation. This approach has yielded little, however, about the genetic basis of reproductive isolation. For example, we do not know yet if species differences are based on many genes of small phenotypic effect, or if few genes of large effect are most important in causing divergence and reproductive isolation. This has made it difficult to pinpoint exactly how natural selection has led to divergence in most cases. Recent studies of speciation in monkeyflowers and other taxa are helping to overcome this gap.The ‘bottom up’ or genetic approach to studying speciation has hunted down genes responsible for premating and postmating isolation, and then shown that the gene sequences exhibit signatures of recent selection. But this approach has told us little about the nature of that selection. Is selection divergent or has divergence occurred under uniform selection? Was selection in response to environmental differences? Was it natural or sexual selection?Finally, we still know little about how mate preferences evolve within and between populations during the process of speciation. Sexual selection by mate choice might be a near-universal process in speciation, but what drives the divergence of mate preferences to begin with?Speciation study is in the midst of a surge of research effort, and part of the reason is that answers to many of these questions appear at last to be within reach. We expect that a combination of phenotypic and genetic approaches will soon close the gap between the genes and the mechanisms of selection, and yield a fuller account of how most species in nature have formed.