Extremely high variability in genes of the major histocompatibility complex (MHC) in vertebrates is assumed to be a consequence of frequency-dependent parasite-driven selection and mate preferences based on promotion of offspring heterozygosity at MHC, or potentially, genome-wide inbreeding avoidance. Where effects have been found, mate choice studies on rodents and other species usually find preference for MHC-dissimilarity in potential partners. Here we critically review studies on MHC-associated mate choice in humans. These are based on three broadly different aspects: (1) odor preferences, (2) facial preferences and (3) actual mate choice surveys. As in animal studies, most odor-based studies demonstrate disassortative preferences, although there is variation in the strength and nature of the effects. In contrast, facial attractiveness research indicates a preference for MHC-similar individuals. Results concerning MHC in actual couples show a bias towards similarity in one study, dissimilarity in two studies and random distribution in several other studies. These vary greatly in sample size and heterogeneity of the sample population, both of which may significantly bias the results. This pattern of mixed results across studies may reflect context-dependent and/or life history sensitive preference expression, in addition to higher level effects arising out of population differences in genetic heterogeneity or cultural and ethnic restrictions on random mating patterns. Factors of special relevance in terms of individual preferences are reproductive status and long- vs. short-term mating context. We discuss the idea that olfactory and visual channels may work in a complementary way (i.e. odor preference for MHC-dissimilarity and visual preference for MHC-similarity) to achieve an optimal level of genetic variability, methodological issues and interesting avenues for further research.