Sex work has been associated with elevated risk for HIV infection among men who have sex with men (MSM) in many settings. This mixed methods study examined sexual risk among MSM sex workers in Massachusetts, collecting formative data on HIV risk behavior by sex worker type in order to gain a better understanding of how to tailor prevention interventions to this unique and high-risk subgroup of MSM. Two groups of MSM sex workers were recruited between January and March 2008: street workers (n=19) and internet escorts (n=13). Participants completed a semistructured qualitative interview and quantitative psychosocial assessment battery; interviews were conducted until redundancy in responses was achieved. Almost one third (31%) were HIV-infected. The majority of participants (69%) reported at least one episode of unprotected serodiscordant anal sex (either insertive or receptive) with a mean of 10.7 (SD=42.2) male sex partners of an unknown or different HIV serostatus in the past 12 months. Salient findings included: (a) internet sex workers reported being paid substantially more for sex than street sex workers; (b) inconsistent condom use, high rates of unprotected sex, and low rates of HIV status disclosure with sex work partners for both internet and street workers; general perceptions of a lack of trust on the part of sex work partners (i.e., telling them what they want to hear), offers of more money for unprotected sex; (c) contextual differences in risk taking: internet sex workers reported that they are more likely to engage in sexual risk-taking with noncommercial sex partners than sex partners who pay; (d) HIV status and STI history: two street workers became infected in the context of sex work, and 25% of the entire sample had never been tested for sexually transmitted infections (STI); and (e) motivations and reasons for doing sex work, such as the "lucrativeness" of sex work, as a means to obtain drugs, excitement, power, "why not?" attitude, and because social norms modeled this behavior. Study findings can be used to generate hypotheses for designing and providing tailored primary and secondary prevention interventions for this at-risk subgroup of MSM.