OBJECTIVES Determining costs of concomitant medications (CONMEDs) is a challenging but critical component of cost-effectiveness analysis. In clinical trials, CONMEDs are typically recorded using broad medication terms, with approximate costs linked to the entire category. In a recent phase III oncology clinical trial, we determined costs with an alternative approach based on the individual CONMEDs used by patients, using a combination of WHO preferred medication term codings, the NDC-HCPCS Crosswalk (CW), and Payment Allowance Limits (PAL) for Medicare Drugs Part. METHODS The CONMED database was obtained from the clinical trial, and the CW and the PAL were obtained from CMS. Preferred medication terms of individual CONMEDs were coded according to the WHODrug version 2003 Q2 dictionary. The CW was used to map preferred medication terms to appropriate HCPCS codes, and the PAL was used to determine unit costs of the HCPCS coded medications. For medications with multiple HCPCs codes, the average payment limit per dosage unit was assigned. Total CONMED costs were computed by adding all cost information. RESULTS The CONMED database comprised approximately 400 patients and 3,588 CONMED records. There were 562 unique HCPCS codes and 497 unique preferred medication terms in the CW and the PAL. In addition, there were 519 unique combinations of preferred medication terms and dose units, of which 78% (407/519) had multiple NDC codes. However, only 17% (70/407) of these had different unit payment limits across products within the combination. In these cases the average cost was used. Overall we were able to assign costs to 22% (780/3588) of CONMED records, consistent with the proportion of CONMEDs covered by Medicare. CONCLUSIONS From third party payer perspective, this micro-costing method for CONMEDs was a feasible approach to pharmacoeconomic assessment with a clinical trial.