When evidence-based programs (EBPs) are scaled up in natural, or non-research, settings, adaptations are commonly made. Given the fidelity-versus-adaptation debate, theoretical rationales have been provided for the pros and cons of adaptations. Yet the basis of this debate is theoretical; thus, empirical evidence is needed to understand the types of adaptations made in natural settings. In the present study, we introduce a taxonomy for understanding adaptations. This taxonomy addresses several aspects of adaptations made to programs including the fit (philosophical or logistical), timing (proactive or reactive), and valence, or the degree to which the adaptations align with the program's goals and theory, (positive, negative, or neutral). Self-reported qualitative data from communities delivering one of ten state-funded EBPs were coded based on the taxonomy constructs; additionally, quantitative data were used to examine the types and reasons for making adaptations under natural conditions. Forty-four percent of respondents reported making adaptations. Adaptations to the procedures, dosage, and content were cited most often. Lack of time, limited resources, and difficulty retaining participants were listed as the most common reasons for making adaptations. Most adaptations were made reactively, as a result of issues of logistical fit, and were not aligned with, or deviated from, the program's goals and theory.