Abstract: Sexual assault imposes substantial social and economic costs, yet existing data likely understate its prevalence. I investigate the decision to report sexual assault and the role of social judgment in suppressing disclosure by both victims and perpetrators. I employ an online experiment with over 1,600 college students and graduates, randomizing respondents to answer questions about rape and other unwanted sexual contact using either direct elicitation or a "hard-garbling" survey technique that provides plausible deniability. The anonymous self-reports used in the experiment mirror the type frequently used by organizations to measure sexual assault prevalence and allow me to abstract away from other barriers to reporting. The experiment also measures perceptions of victim-focused social norms. I find widespread beliefs that victims may be lying and vast misperceptions about the prevalence of victim-blaming, suggesting that victims expect significant negative judgment for disclosing. Importantly, plausible deniability substantially increases disclosure. Among women, hard-garbling increases reports of non-rape sexual assault victimization by 23 percent. Among men, it more than doubles reporting of rape perpetration: 13.1 percent of men admit to having raped someone under hard-garbling, relative to 6.2 percent in direct elicitation. My findings suggest that sexual assault experiences remain under-reported even in anonymous surveys and provide strong evidence that concerns about social judgment play an important role in suppressing reports of sexual assault by both victims and perpetrators.