Based on data from the VIRAGE (Violence and Gender Relations: INED, 2015–2016) survey, this article explores reports of violence inminority populations through the case of family violence reported by members of sexual minorities. VIRAGE provides two samples of homo-/bisexual respondents who answered the same questionnaire: a volunteer sample recruited through a communications campaign, who responded via Internet; and respondents to a general population telephone survey who identified as lesbian, gay or bisexual. Sexual identification is a sensitive issue, and surveys of private households do not necessarily capture either all victims of violence or all of those who identify as homosexual or bisexual. The general population sample thus cannot be considered representative of these groups. Lesbian and bisexual women in both samples reported more family violence than gay and bisexual men. Homo-/bisexual respondents in VIRAGE’s convenience sample reported more experiences of violence than those in its general population sample. This difference may be explained by various factors: effects of the data collection mode, and of data collection strategies more generally, as well as the heterogeneity of populations in terms of either exposure to violence or the propensity to report it. We use logistic regression methods to understand these differences, controlling for the differences between the characteristics of the respondents in the two samples. In addition to sociodemographic characteristics, we hypothesized that sexual trajectories and practices of sociability among homosexuals and bisexuals could affect the reporting and perception of violence. After these analyses, differences remain, but to a variable extent depending on sex, sexual identification, and forms of violence. The resulting models explain the differences in reporting between samples for men less well than for women. This result can be interpreted as the consequence of a difference between women’s and men’s exposure to violence: independently of the effects of data collection mode and sampling differences, the high probability that female sexual minorities will experience family violence leads to relatively high reporting in the two samples.