Specialty-related cost differences for the treatment of nonmelanoma skin cancer (NMSC) have been previously reported but without taking into account confounding factors. Using a previously validated model for NMSC episode of care, episodes were identified in the Medicare Current Beneficiary Survey claims 2005 to 2007. A γ regression with log link model estimated the effect of physician exposure on total episode costs controlling for sociodemographics, health status and comorbidities, treatment and repair procedures, as well as tumor size and location. Treatment-related NMSC episodes (1285) were identified. In the unadjusted model, episodes managed by generalists were associated with 36% lower costs, those by otolaryngologists/plastic surgeons with 82% higher costs, and those by multiple specialists with 111% higher costs, compared to dermatologists. Cost differences were substantially reduced in the adjusted regression analysis; compared to dermatologists, episodes managed by generalists were associated with 20% lower costs (P < 0.0001), whereas otolaryngologists/plastic surgeons and multiple specialists were associated with 20% (P < 0.01) and 11% (P = 0.02) higher costs, respectively. Overall, comparison between unadjusted and adjusted estimates suggests that controlling for severity and treatment modalities explains most of the specialty cost differences. Our estimates could be subject to residual confounding due to selection bias and the limitations to using claims data to characterize an NMSC episode of care. Adjusting for the severity of the disease and other confounders, our study found much smaller specialty-related cost differences for the management of NMSC than previously reported unadjusted estimates.