OBJECTIVES: To evaluate short-term and long-term cost-effectiveness of imiquimod versus surgery in the treatment of superficial basal cell carcinoma in the Netherlands. METHODS: A decision analytic model adopting a societal perspective was developed and compared cost and outcomes of imiquimod vs. surgery. The short-term (18 weeks) effectiveness outcome was histological clearance, with sustained clearance the outcome at 3 years. Direct costs included costs of excision and Moh's surgery, drugs, adverse events, follow-up and transportation. Indirect costs comprise working hours lost due to dermatologist and surgery visits. Data were derived from clinical trials (Imiquimod histological clearance and recurrence rates), Delphi panel (resource utilisation) and published literature (surgery response and recurrence rates). Cost data were taken from official costing guide and tariff lists. Two scenarios were used for calculating surgery costs: 1) micro-costing (using average time and supplies obtained from the Delphi panel) and 2) a Dutch study on the costs of surgery. Long-term costs were discounted at 4%. RESULTS: Compared with surgery, short-term savings with imiquimod were €79 and €97 per patient for scenarios 1 and 2 respectively [total costs: €585 vs. €663 and €590 vs. €687]. Histological clearance of 82% for imiquimod and 91% for surgery made imiquimod a cost-effective treatment option. Long-term costs with imiquimod were an extra €148 and €133 for scenarios 1 and 2 [total costs: €1,471 vs. €1,322 and €1,479 vs. €1,346], while sustained clearance with imiquimod was 87% vs. 96% for surgery. The model showed moderate sensitivity to changes in response and recurrence rates, response to treatment, and numbers of working days lost. CONCLUSION: While imiquimod is cost-effective in the short-term, long-term cost-effectiveness should be judged with prudence because of the uncertainty surrounding long-term recurrence data. Non-tangible benefits, such as patient preference for avoiding surgery and patient convenience were not quantified in the model either.