The utility of mathematical models in understanding the dynamics of HIV transmission among injecting drug users (IDUs) and their non-IDU sex partners is discussed. We emphasize the need for collaborative relationships between modellers and drug-use researchers, and we stress that models should be based on data in both their formulation and development stages. We outline some of the possible data requirements of transmission models and we highlight the need for the collection of appropriate quantitative data, so that modellers can estimate specific parameters for their models. We believe that discussion of the needs and utility of mathematical modelling will lead to mutually beneficial collaborations between theoreticians and drug-use researchers. These collaborations may aid in the design and the evaluation of effective behavioural or medical intervention strategies.