The purpose of this study was to describe and compare the rates of recruitment during a randomized clinical trial on smoking cessation in two primary care practices. One site was a five-physician private family practice setting with about 15,000 patients. During 34 days, 576 patients were screened, of whom 22% were smokers. Among the smokers screened, 54% consented, 33% refused consent, and 13% were called in too early to consent. The other site was a six-physician academic medical practice with about 16,000 patients. During 53 days, 1,692 subjects were screened, of whom 16.2% were smokers. Among the smokers, 19% consented, 81% refused consent, and none were called in early. The enrollment of smokers was 3.3 times greater in the private practice than the academic practice. At the first site, study personnel screened 26.6 subjects per day, whereas the practice receptionist screened only 13.4 subjects per day (P less than .01). A randomized trial of having subjects read the informed consent versus having study personnel read it to them showed no differences in recruitment. The data suggest that private practices may have greater potential for subject recruitment than academic sites, that using study personnel improves recruitment, and that having study personnel actively involved in informed consent does not improve recruitment.