The sample included 84 IDU recruited in Sydney, Australia. A computer game-like format was used to generate hypothetical injecting events. Levels of situational factors were varied randomly between situations. These included needle availability, bleach availability, craving, how well known potential sharing partners were, previous sharing, privacy of site, and the degree to which individuals were drunk or stoned. Responses to choices in each injecting event included the likelihood of sharing in each situation, the frequency with which they experienced the hypothetical situation in real life, and satisfaction with their response. Canonical correlation analysis indicated the association between situational influences and choice variables accounted for 87.78 percent of the variance. Sharing needles was negatively related to decision satisfaction and experience with an actual situation, indicating that IDU internalized HIV preventive norms and were not likely to share on a normal basis. Choices about needle sharing were primarily constrained by the tradeoff between needle availability and craving, while social factors played a secondary role. The probability of sharing increased as other people present were better known and previous sharing history increased. This pattern of results suggest that, while needle exchanges can be effective, they may fail to reach their full potential unless factors related to substance use and social context are also addressed. 29 references, 1 table, and 1 figure