Background: The p53 tumor suppressor is frequently mutated in colon cancer, but the influence of such mutations on survival is remains undefined. We investigated whether domain-specific mutations in p53 are predictive of survival in stage III colon cancer. Methods: p53 was evaluated in an intergroup trial (CALGB 89803) of patients with stage III colon cancer who were randomized to receive adjuvant 5-fluorouracil/leucovorin (5FU/LV) or 5FU/LV with irinotecan (IFL) Tissue was collected to allow correlation of molecular markers with outcomes. p53 was genotyped in 607 patient tumors. Results: p53 mutations were identified in 274 tumors, divided ~ equally between zinc binding and non-zinc binding regions of the DNA binding domain. Overall, p53 status was not predictive of benefit from either adjuvant regimen. Unexpectedly, the 5 year overall survival (OS) of women with tumors harboring non-zinc binding mutations treated with 5FU/LV was 97% compared to OS of 72% for women with p53 wild-type (wt) tumors (p =0.004). Adding irinotecan to 5FU/LV negated this survival benefit (5 year OS of 81% vs. 72%). Conversely, 5 year OS of women harboring tumors with zinc binding mutations who received 5FU/LV was 50% compared to 72% for women with p53 wt tumors (p=0.04). Adding irinotecan to 5FU/LV reversed the poor survival of women with tumors harboring zinc binding mutations and improved 5 year OS (50% vs. 73%; p=0.1). No difference in OS was observed for men on either treatment arm or when genotype was considered. Conclusions: CALGB 89803 demonstrated a lack of survival benefit for stage III colon cancer patients when irinotecan was added to 5FU/LV (IFL). We now show that in the setting of a large clinical trial, refined stratification of women, based upon domain- specific mutations of p53 identifies subsets of patients likely to benefit from, or respond poorly to, adjuvant 5FU/LV. The interaction of p53 genotype, gender, and adjuvant therapy regimen has the potential to be paradigm changing in the treatment of colon cancer, and possibly other malignancies. These data, if validated, suggest that evaluation of p53 genotype and gender may guide clinicians to make rational choices of adjuvant therapy.