Scholars of nonprofits, interest groups, civic associations, and social movement organizations employ samples of organizations derived from directories or other available listings. In most cases, we are unable to evaluate the representativeness of these samples. Using data on the population of environmental organizations in North Carolina, we assess the methodological strengths and weaknesses of widely used strategies. We find that reliance on any single source yields bias on theoretically important characteristics of organizations. We show that scholars can reduce bias significantly by combining sources, creating what we call a ‘peak list’ compiled from different types of sources. Compared to any single source, our peak list differed less from the population on the thirty-one organizational characteristics including geographical coverage, issues, discursive frames, targets, and organizational demographics such as age, organizational form, and resources. From these analyses, we offer methodological recommendations for making better-informed decisions for constructing representative organizational samples.