Abstract Recent studies of somatic and germline mutations have led to the identification of a number of factors that influence point mutation rates, including CpG methylation, expression levels, replication timing and GC content. Intriguingly, some of the effects appear to differ between soma and germline: in particular, whereas mutation rates have been reported to decrease with expression levels in tumors, no clear effect has been detected in the germline. Distinct approaches were taken to analyze the data, however, so it is hard to know whether these apparent differences are real. To enable a cleaner comparison, we considered a statistical model in which the mutation rate of a coding region is predicted by GC content, expression levels, replication timing, and two histone repressive marks. We applied this model to both a set of germline mutations identified in exomes and to exonic somatic mutations in four types of tumors. Germline and soma share most determinants of mutations; notably, we detected an effect of expression levels on germline mutations as well as on somatic ones. However, whereas in somatic tissues, increased expression levels are associated with greater strand asymmetry and decreased mutation rates, in ovaries and testes, increased expression leads to greater strand asymmetry but increased mutation rates. This contrast points to differences in damage or repair rates during transcription in soma and germline.