Randomized controlled trials are the gold standard in research but may not fully explain or predict outcome variations in community-based interventions. Demonstrating efficacy of externally driven programs in well-controlled environments may not translate to community-based implementation where resources and priorities vary. A bottom-up evaluation approach, in which community-developed interventions are evaluated for efficacy and then disseminated, improves the relevance of findings for real-world settings. The Girlfriends Project is an HIV risk reduction intervention developed by and for African American women, which I evaluated using a bottom-up approach. In this article, I discuss methods to increase community engagement and evaluation rigor, including retaining community members to assist with participant recruitment and data collection, as well as facilitating a shared decision-making process between the evaluator and community members. I also discuss specific challenges and strategies for the evaluation of community-developed interventions in real-world settings and highlight the value of this approach.