Cybersex (engaging in sexual self-stimulation while online with another person) and meeting partners for sex are commonly reported among men who have Internet sex with men (MISM). We studied 1,026 Latino MISM in the US and constructed a scale to measure beliefs about cybersex. Respondents were recruited to an online Internet study and compensated for their effort. Factor analysis indicated that there were three dimensions: (1) liking cybersex: perceived anonymity and safety of cybersex; (2) potential for greater IRL interaction in seeking partners on the Internet; and (3) a dimension of negative beliefs about Internet contacts which was not psychometrically stable. Safety referred to physical and HIV-related safety. We formed two scales to describe the first two dimensions. Respondents who were high on the 'liking cybersex: perceived anonymity and safety' scale were more likely than low scorers to use chat-rooms and webcams. They were also more likely than low scorers to have had cybersex and phone sex before meeting. Those who were high on the 'IRL interaction' scale were more likely than low scorers to use e-mail, bulletin boards and paid websites. However, there was significantly less disclosure about HIV status and discussion about safer sex by those higher on the 'IRL interaction scale', and when sex in real life did occur, those higher on the 'IRL interaction' scale were more likely to have receptive oral or anal sex, whereas those high on the 'liking cybersex: perceived anonymity/safety' scale were less likely to have insertive sex or be drunk/high. MISM scoring high on the 'IRL interaction' scale were significantly more likely than low scorers to prefer to meet partners on the Internet. We discuss the implications of the Internet for MSM and for sexual IRL interactions in the Internet age.