There is increasing recognition of the value from incorporating local and scientific hazard knowledge for disaster risk management, but practical integration remains limited. Much management activity remains underpinned by information deficit thinking, which inherently valorizes scientific knowledge and devalorizes local knowledge. To move away from deficit thinking, we develop the concept of “cross-validation” as a step toward integrating types of hazard knowledge. Rather than science validating local knowledge, cross-validation is multidirectional and focuses on identifying the strengths of each knowledge type and their areas of complementarity, to achieve a more complete hazard understanding. We applied the cross-validation approach to analyze geohazard knowledge in two case study sites in southern Peru. We identified substantial agreement around the most prominent local hazards, providing mutual reinforcement of each type of knowledge. We also identified complementary facets, where strengths in one knowledge type compensated for limitations in the other. The cross-validation approach shows promise as a first step toward incorporating knowledge, providing an opportunity to establish the foundation from which practical integration can build.