Medical investigators increasingly attempt to demonstrate the effectiveness of treatments given in routine practice settings as well as in research settings. We identified and describe several key issues in practice-based research: identifying the appropriate population; drawing the sample; recruitment of physicians and patients; maintenance of the physician-researcher relationship; and data management. Three on-going practice-based projects are described from the perspective of these key issues. Two are practice-based randomized trials, one is a larger cohort study. Common characteristics are identified which contributed to the success of the three projects: direct recruitment of clinicians by clinicians; close liaison with local medical organizations; on-going personal contact with the practices; and recognition of the value of the practicing physician's time.