4009 Purpose: To examine 4 possible prognostic factors in patients (pts) with recurrent (rec) stage II and III colon cancer: time from randomization to rec (< 1 yr, 1–2 yrs 2–3 yrs 3–4 yrs, 4+ yrs), initial stage (II vs III), initial adjuvant treatment (adj rx) (5-FU based vs no rx control), and era pt entered onto study (1978–1985, 1986–1992, 1993–1999). Methods: The Adjuvant Colon Cancer Endpoints (ACCENT) dataset, including individual pt data from 18 trials and over 20,800 pts testing 5-FU based adj rx for pts with stage II and III colon cancer, was analyzed using univariate and multivariate Cox proportional hazards models, stratified by study. The QUASAR trial (Lancet, 2000) was excluded due to non protocol specified follow-up for rec. No data was available on rx subsequent to rec, or tumor characteristics at time of rec. Results: 5,713 of 17,381 (32.9%) of pts experienced rec. Median survival following rec was 13.4 months. Primary results from univariate models are shown in the table . Time from randomization to rec was highly prognostic of survival following rec (p<0.0001); this result was primarily driven by pts with initial stage III disease. Longer survival following rec was also seen in pts with initial stage II vs III disease (p<0.0001), with a 15% 6 year OS in initial stage II pts, pts entered more recently onto trials (p< 0.0001), and pts initially treated with surgery alone vs 5-FU adj rx (p=0.0005). All relationships were maintained in multivariate models. Conclusions: Time from initial rx and initial stage are important prognostic factors in pts with rec colon cancer. Survival following rec increased modestly from 1980–1995 possibly due to improvements in surgery, chemotherapy, supportive care, or unknown selection effects. Pts who rec following adj rx have poorer prognosis than those untreated, which may be a selection effect (only more aggressive tumors escape adj rx), or a sensitivity effect (those without adj rx retain sensitivity to 5-FU in the rec setting). [Table: see text] No significant financial relationships to disclose.