The transmission of infectious diseases is an inherently ecological process involving interactions among at least two, and often many, species. Not surprisingly, then, the species diversity of ecological communities can potentially affect the prevalence of infectious diseases. Although a number of studies have now identified effects of diversity on disease prevalence, the mechanisms underlying these effects remain unclear in many cases. Starting with simple epidemiological models, we describe a suite of mechanisms through which diversity could increase or decrease disease risk, and illustrate the potential applicability of these mechanisms for both vector-borne and non-vector-borne diseases, and for both specialist and generalist pathogens. We review examples of how these mechanisms may operate in specific disease systems. Because the effects of diversity on multi-host disease systems have been the subject of much recent research and controversy, we describe several recent efforts to delineate under what general conditions host diversity should increase or decrease disease prevalence, and illustrate these with examples. Both models and literature reviews suggest that high host diversity is more likely to decrease than increase disease risk. Reduced disease risk with increasing host diversity is especially likely when pathogen transmission is frequency-dependent, and when pathogen transmission is greater within species than between species, particularly when the most competent hosts are also relatively abundant and widespread. We conclude by identifying focal areas for future research, including (1) describing patterns of change in disease risk with changing diversity; (2) identifying the mechanisms responsible for observed changes in risk; (3) clarifying additional mechanisms in a wider range of epidemiological models; and (4) experimentally manipulating disease systems to assess the impact of proposed mechanisms.