The function of meiotic recombination has remained controversial, despite recent inroads into mechanisms. Ideas concerning a possible role of recombination in the elimination or efficient incorporation of mutations have been backed by theoretical studies but have lacked empirical support. Recent investigations into the basis for local variations in recombination frequency in yeast have uncovered a strong association between recombination initiation sites and transcriptional regulatory sequences. Other recent studies indicate a strong correlation between transcription and mutation rates in yeast genes. Taken together, these data imply that distributions of recombination and mutation frequencies may be strongly correlated. This suggests that recombination may be targeted to genomic sites of high mutation frequency; such a 'mutation-tracking' function would clearly aid in the shuffling of mutations to break up unfavorable and create favorable allelic combinations. Moreover, recent insights into the mechanism of gene conversion in yeast reveal a very strong inherent bias in favor of alleles on the non-initiating homolog. Combined with mutation tracking, these findings suggest a novel and general mechanism by which allelic gene conversion may act to eliminate mutations.