Background: Tobacco cessation interventions by physicians hold promise in improving quit rates. The 5As intervention ('Ask', 'Advise', 'Assess', 'Assist' and 'Arrange') is an evidence-based approach for tobacco cessation. However, little is known about adherence with the tobacco cessation interventions in primary care in India. In the present study we assessed physicians' adherence with the 5As intervention and explored physician and patient concordance on the report of 5As intervention for tobacco cessation. Methods: We used data from two cross-sectional surveys conducted in 12 districts of Andhra Pradesh and Gujarat in India. The surveys were administered simultaneously to both patients attending, and physicians working in health facilities providing primary care. Health facilities were selected by systematic random sampling and patients were recruited by simple random sampling. Common health facilities where both surveys were performed were identified, and individual patients were matched to their physicians through a unique matching code to obtain the two study samples. Results: Slight agreement was observed between the physician and patient responses on 'Ask' and 'Arrange' component of the 5As intervention. The 'Advise', 'Assess' and 'Assist' components showed low agreement. Slightly higher levels of agreement were seen on all components of the 5As, except 'Advise', for those patients who had made an attempt to quit. Conclusions: Our study suggests an urgent need for revising current strategies in order to strengthen the 'Advise', 'Assess', and 'Assist' interventions in tobacco cessation in primary care settings. Patient surveys should be used routinely in assessing fidelity and provider adherence for large scale behavioral health programs.