Scientific understanding of the role of development in conservation has been hindered by the quality of evaluations of integrated conservation and development projects. We used a quasi-experimental design to quantitatively assess a conservation and development project involving commercial butterfly farming in the East Usambara Mountains of Tanzania. Using a survey of conservation attitudes, beliefs, knowledge, and behavior, we compared 150 butterfly farmers with a control group of 170 fellow community members. Due to the nonrandom assignment of individuals to the two groups, we used propensity-score matching and weighting in our analyses to control for observed bias. Eighty percent of the farmers believed butterfly farming would be impossible if local forests were cleared, and butterfly farmers reported significantly more participation in forest conservation behaviors and were more likely to believe that conservation behaviors were effective. The two groups did not differ in terms of their general conservation attitudes, attitudes toward conservation officials, or knowledge of conservation-friendly building techniques. The relationship between butterfly farming and conservation behavior was mediated by dependency on butterfly farming income. Assuming unobserved bias played a limited role, our findings suggest that participation in butterfly farming increased participation in conservation behaviors among project participants because farmers perceive a link between earnings from butterfly farming and forest conservation.