Objective: There is a substantial body of research on risk factors for violent behavior in adulthood but little empirical study of protective factors and desistance. Method: This study aimed to investigate the protective effects of factors hypothesized to reduce violent reoffending among a sample of 800 male prisoners following release into the community. Results: Five out of 15 hypothesized protective factors significantly reduced the likelihood of a violent reconviction. Protection was primarily related to social network factors that appeared to have long-term effects. Circumstances following release (finances and nature of accommodation) did not prove to be relevant, apart from immediate provision of accommodation, which had a substantial influence on reduction of risk shortly after release into the community, and being employed, in training or education, which was relevant for the 3rd year and subsequently. Our findings support the notion of independent effects of protective factors on outcome. No effect modification was found to indicate that protective factors have different effects at different levels of risk. Conclusions: The nature of the relationship between risk and protection should be considered in the assessment of risk for violence. Furthermore, risk factors focus on pathology and hazards, whereas protective factors emphasize positivism and hope for change. Future interventions may benefit from this positive approach. (Contains 2 tables.)