Objective: To explore the cost-effectiveness of capecitabine as adjuvant treatment for Stage III colon cancer. Phase III clinical trials show that capecitabine improves disease-free survival. However, these trials involved younger patients than reflected clinically and overall survival was not significantly better than with usual care. We conducted a modeling study comparing the cost-effectiveness of capecitabine and standard care (Fluorouracil/Leucovorin (5FU/LV)) in a public-payer context (Canada), using an older cohort, and with overall survival as the main outcome. Methods: A Markov model was developed to determine the cost-effectiveness of capecitabine compared with 5FU/LV. The base case was a 70-year-old man after total mesorectal resection excision of Stage III colon cancer. A five year time horizon was used. Health states included treatment phase, remission, recurrence, disease progression, and death; throughout the model (except during the active treatment states) patients could die from other risk-related causes. Ontario health economic data were used for costs. Probabilities were obtained from the published literature, and sensitivity analyses were conducted. Results: The base case costs for capecitabine and 5FU/LV were $12,999 and $12,191, respectively. Overall survival was 4.132 and 4.069 years, respectively. The incremental cost-effectiveness ratio of capecitabine was $12,821 per life year gained. However, the incremental cost-effectiveness ratio of capecitabine was greater than $50,000/life year when the annual probability of relapse was greater than 0.96 or when drug costs were assumed to be greater than $1410 per cycle (both values within the plausible range). Conclusion: Capecitabine produced modestly improved survival over 5FU/LV (0.063 extra years) with a favourable cost-effectiveness ratio. However, because the model was sensitive to variations in relapse rate and drug costs, the relative attractiveness of capecitabine over 5FU/LV is not certain. In addition, utilities and indirect costs were not considered in the model. Because capecitabine is administered orally, this could be an important factor warranting further research.