Objectives (1) To characterise variation in general practitioners' (GPs') accounts of communicating with men about prostate cancer screening using the prostate-specific antigen (PSA) test, (2) to characterise GPs' reasons for communicating as they do and (3) to explain why and under what conditions GP communication approaches vary. Study design and setting A grounded theory study. We interviewed 69 GPs consulting in primary care practices in Australia (n=40) and the UK (n=29). Results GPs explained their communication practices in relation to their primary goals. In Australia, three different communication goals were reported: to encourage asymptomatic men to either have a PSA test, or not test, or alternatively, to support men to make their own decision. As well as having different primary goals, GPs aimed to provide different information (from comprehensive to strongly filtered) and to support men to develop different kinds of understanding, from population-level to 'gist' understanding. Taking into account these three dimensions (goals, information, understanding) and building on Entwistle et al's Consider an Offer framework, we derived four overarching approaches to communication: Be screened, Do not be screened, Analyse and choose, and As you wish. We also describe ways in which situational and relational factors influenced GPs' preferred communication approach. Conclusion GPs' reported approach to communicating about prostate cancer screening varies according to three dimensions-their primary goal, information provision preference and understanding sought-and in response to specific practice situations. If GP communication about PSA screening is to become more standardised in Australia, it is likely that each of these dimensions will require attention in policy and practice support interventions.