Background Recent evidence suggests that the majority of HIV transmissions among men who have sex with men (MSM) occur between steady partners. We sought to determine factors associated with HIV transmission risks in steady partnerships. Methods Data is from the German cross-sectional 2013 Gay Men and AIDS survey. The study population was HIV-negative or untested men reporting a steady partnership and at least one non-steady anal sex partner in the previous year. Bivariate and multivariate logistic regression was used to determine which of several independent variables best predicted both unprotected anal intercourse (UAI) with a non-steady partner and lack of HIV testing in the past year (high-risk outcome group). Results The study population consisted of 1731 men. Among individuals in the outcome group (n = 271), 67 % reported UAI with a non-steady partner of unknown status and 9 % reported UAI with a non-steady HIV-positive partner in the past 12 months; 55 % considered themselves to be at low risk for HIV acquisition. In multivariate analyses (n = 1304), participants were statistically more likely to belong to the outcome group if they reported UAI with their steady partner in the past year (OR = 2.21), did not know their steady partner’s HIV status (OR = 1.98), or agreed that condoms were disruptive during sex (OR = 3.82 (strongly agree), OR = 2.19 (agree)). Participants were less likely to belong to the outcome group if they were out to their primary doctor (OR = 0.54), were well-educated about post-exposure prophylaxis (OR = 0.46), had sought information on HIV in the past year and kept condoms in an accessible place (OR = 0.20), or believed that insisting on condoms would lead partners to assume they were HIV-negative (OR = 0.20). Participants in the outcome group were more likely to say they would use HIV home tests (OR = 1.58) or pre-exposure prophylaxis (OR = 2.11). Conclusions Based on our results, we reflect on HIV prevention measures that should be improved in order to better target behaviors that may lead to HIV transmission between MSM in steady relationships. In particular, we highlight the need for multifaceted interventions focusing not only on members of the at-risk community themselves, but on communities as a whole.