Background: We compared two methods of measuring provider performance of tobacco control activities: immediate "exit cards" versus delayed telephone follow-up surveys. Current standards, e.g. HEDIS, use delayed patient measures that may over or under-estimate overall performance. Methods: Patients completed exit cards in 60 dental practices immediately after a visit to measure whether the provider "asked" about tobacco use, and "advised" the patient to quit. One to six months later patients were asked the same questions by telephone survey. Using the exit cards as the standard, we quantified performance and calculated sensitivity (agreement of those responding yes on telephone surveys compared with exit cards) and specificity (agreement of those responding no) of the delayed measurement. Results: Among 150 patients, 21% reporting being asked about tobacco use on the exit cards and 30% reporting being asked in the delayed surveys. The sensitivity and specificity were 50% and 75%, respectively. Similarly, among 182 tobacco users, 38% reported being advised to quit on the exit cards and this increased to 51% on the delayed surveys. The sensitivity and specificity were 75% and 64%, respectively. Increasing the delay from the visit to the telephone survey resulted in increasing disagreement. Conclusion: Patient reports differed considerably in immediate versus delayed measures. These results have important implications because they suggest that our delayed measures may over-estimate performance. The immediate exit cards should be included in the armamentarium of tools for measuring providers' performance of tobacco control, and perhaps other service delivery. © 2008 Houston et al; licensee BioMed Central Ltd.