Fieldwork is a critical tool for scientific research, particularly in applied disciplines. Yet fieldwork is often unsafe, especially for members of historically marginalized groups and people whose presence in scientific spaces threatens traditional hierarchies of power, authority, and legitimacy. Research is needed to identify interventions that prevent sexual harassment and assault from occurring in the first place. We conducted a quasi-experiment assessing the impacts of a 90-min interactive training on field-based staff in a United States state government agency. We hypothesized that the knowledge-based interventions, social modeling, and mastery experiences included in the training would increase participants’ sexual harassment and assault prevention knowledge, self-efficacy, behavioural intention, and behaviour after the training compared to a control group of their peers. Treatment–control and pre-post training survey data indicate that the training increased participants’ sexual harassment and assault prevention knowledge and prevention self-efficacy, and, to a lesser extent, behavioural intention. These increases persisted several months after the training for knowledge and self-efficacy. While we did not detect differences in the effect of the training for different groups, interestingly, post-hoc tests indicated that women and members of underrepresented racial groups generally scored lower compared to male and white respondents, suggesting that these groups self-assess their own capabilities differently. Finally, participants’ likelihood to report incidents increased after the training but institutional reports remained low, emphasizing the importance of efforts to transform reporting systems and develop better methods to measure bystander actions. These results support the utility of a peer-led interactive intervention for improving workplace culture and safety in scientific fieldwork settings.