Abstract As part of a large sexual health study, we used two different approaches to target Sexual Minority Individuals (SMIs). Firstly, we drew on a probability sample (1,832 respondents aged 14-80) of the Flemish population in Belgium. Secondly, we set up a targeted sampling design followed by an Internet survey. Our focus was to explore how two different sampling procedures and survey designs could lead to differences in sample characteristics. Results showed that for female SMIs (we excluded male SMIs from the analyses due to their low numbers) the population sample differed from the Internet sample in terms of sociodemographic characteristics (the latter included younger and more highly educated respondents) and scores on sexual orientation dimensions (the population sample included more respondents who didn’t identify as lesbian or bisexual but reported same-sex sexual experiences and desire). Respondents’ scores on sexual health indicators differed between the samples for two of the seven variables. We discuss implications for improving the quality and validity of nonrandom samples.