Recruitment and Enrollment Assessment in Clinical Trials (REACT), an NHLBI-sponsored substudy of the Cardiac Arrhythmia Suppression Trial (CAST), was conducted to assess factors associated with enrollment in clinical trials. We report on the relationships of institutional factors at CAST sites to patient enrollment. The proportion of CAST-eligible patients enrolling at each CAST site during the REACT study period was defined as the number of subjects enrolled divided by the sum of (1) the number enrolled plus (2) the number of eligibles who refused plus (3) the number of eligibles whose physicians refused to permit CAST personnel to attempt to enroll them. A questionnaire that included 78 questions regarding factors hypothesized to be associated with enrollment was completed between August 1988 and February 1990 by the nurse coordinators at all 112 CAST sites in the United States and Canada. Sixteen items were unanalyzable, and 37 of the remaining 62 were grouped into seven scales. The remaining items were analyzed individually. Enrollment proportions varied widely across the 112 CAST sites (mean 32.7% SD 22.6). Five variables or scales were included in the final multiple regression model (multiple R2 = .39). The most important of these was the proportion of eligible patients at a site cared for by medical staff other than private attending physicians (multiple R2 for this variable alone, .26). This proportion tended to be high in teaching hospitals. Other variables in this model that were associated with higher enrollment proportions included the number of days per week a nurse coordinator was present at the site, the number of nurse coordinator full-time equivalents at the site, fewer other clinical trials for which the nurse coordinator was responsible, and fewer perceived obstacles to enrollment. These findings indicate that enrollment was more successful at hospitals with higher proportions of eligible subjects cared for by fellows, housestaff, and service attending physicians and at institutions with the committed presence of a nurse-coordinator.