Research on the epidemiology of HIV infection among IV drug users is still at a relatively early stage. Multilocation studies that would permit better geographic comparisons are greatly needed. Multi-method studies within single geographic areas are also needed to assess possible biases with respect to sample recruitment and data collection procedures. The continuation of the epidemic provides a changing historical context that complicates any comparisons. Despite these problems, there are some consistencies that can be seen across studies. Studies of HIV seroprevalence among IV drug users show wide variation among cities in the United States and Europe. The time that the virus was introduced into the IV drug using group within the city is one factor in explaining these differences; other cross-city factors have yet to be identified. Once HIV has been introduced into the IV drug use group within a particular geographic area, there is the possibility of rapid spread up to seroprevalence levels of 50% or greater. Thus, a currently low seroprevalence rate should not be seen as a stable situation. Frequency of injection and sharing of equipment with multiple other drug users (particularly at shooting galleries) have been frequently associated with HIV exposure. Being female, ethnicity (in the USA) and engaging in prostitution also may be associated with increased risk for HIV exposure, suggesting that prevention programs should include special consideration of sex and ethnic differences. Studies of AIDS risk reduction show that substantial proportions of IV drug users are changing their behavior to avoid exposure to HIV. This risk reduction is probably more advanced in New York, with its high seroprevalence and incidence of cases, but is also occurring in cities with lower seroprevalence and limited numbers of cases. The primary forms of risk reduction are increasing the use of sterile equipment, reducing the number of needle sharing partners, and reducing the frequency of injection. These behavior changes are very similar to the frequently identified behavioral risk factors associated with HIV exposure, suggesting that they should be effective in at least slowing the spread of HIV among IV drug users. No linkage of risk reduction to decreases in seroconversion has yet been shown, however, and greater risk reduction is clearly required. A variety of prevention strategies will probably be needed to reduce the spread of HIV among IV drug users. Prevention of initiation into drug injection is an undeniable long-term goal for the control of HIV infection, but there is very little research being conducted in this area.