OBJECTIVES: The objectives of this study are to estimate the diffusion curve of atypical antipsychotic agents and to investigate the influence of physician mix and supply on the diffusion cycle. METHODS: Retrospective analysis of Iowa Medicaid pharmacy claims data were done for the period January 1990 till June 2000 to identify the percentage of patients receiving atypical agents in each county and each month. Antipsychotic agents were identified using the National Drug Codes for antipsychotic agents. Information on physicians practicing in each county was obtained from the Iowa Physician Database. To test whether the diffusion cycle follows the S-shaped curve, a cubic regression model was run where the dependent variable is percentage of patients receiving atypical agents in each month and the independent variable is time. The effect of physician factors were tested using weighted multiple regression analysis where each independent variable was interacted with time, time2 and time3. RESULTS: The number of observations to test the shape of the diffusion curve is 126 (state aggregate over sample period) and the number of observations to test the influence of physician mix and supply is 12,474 (99 counties over sample period). The results revealed that time and time3 were negatively significant and time2 was positively significant suggesting that the diffusion curve resembles the S-shaped curve. Diffusion of atypical agents accelerated earlier in counties having higher number of physicians. Diffusion accelerated faster in counties having higher percentage of high network physicians and counties having higher percentage of younger physicians. Information available was not sufficient to indicate that high psychiatrist counties had a faster diffusion of atypical agents. CONCLUSIONS: Physician mix and supply factors affected differently at different phases of the diffusion cycle. Understanding the role of physician mix and supply on uneven diffusion will help the policy makers develop interventions to take care of this disparity.