OBJECTIVE: Smoking in pregnancy is the foremost cause of preventable perinatal mortality. We have demonstrated that a behavioral intervention can alter smoking in pregnant women. We tested the utility of this intervention at multiple sites in varied settings across a suburban-rural state. STUDY DESIGN: We performed a prospective cohort study at 10 prenatal care sites across North Carolina. Carbon monoxide manometry was used to verify cessation; self-report confirmed reduction. Each site enrolled smokers for 1 year. Four outcome predictor variables were studied: clinic volume, prevalence of smoking, physician versus nonphysician intervenors, and public versus private clinics. RESULTS: Smoking prevalence varied from 4% to 85%. Biologically confirmed quit rates ranged from 0% to 45%. The prevalence of smoking within a clinic's population was able to explain differences in reduction (p < 0.01) of smoking between sites. CONCLUSION: We have demonstrated the effectiveness of an intervention to alter smoking behavior in pregnancy. It appears that this technique has the greatest utility in clinics with a high prevalence of smoking.