BACKGROUND This study evaluated the activity of 2 schedules of erlotinib in combination with chemotherapy, and the prognostic significance of serum amphiregulin (AREG) and transforming growth factor alpha (TGFa) in metastatic colorectal cancer. METHODS A total of 60 untreated patients were randomized to a "continuous" (CON; erlotinib 100 mg daily) or an " intermittent" (INT; erlotinib 150 mg on alternate day on day 2 to 14, then 150 mg daily on days 15 to 21) schedule of erlotinib with a modified XELOX (capecitabine plus oxaliplatin) regimen. Serum levels of AREG and TGFa were determined serially. RESULTS Baseline characteristics were similar between the 2 arms. Of the 58 patients evaluated for response, there was a nonsignificant trend toward a slightly higher overall response rate in the INT arm (66.7%) versus the CON arm (56.7%). At a median follow-up of 2.8 years, the median overall survival was 18.8 months (95% confidence interval = 11.3-22.9 months) and 20.7 months (95% confidence interval = 12.5-31 months, P =.19) for the CON and INT arm, respectively. KRAS mutation did not predict drug response. The 2 arms did not differ significantly in toxicity. Baseline serum TGFa was an independent predictor of progression-free survival, whereas a drop in serum TGFa and AREG levels following 3 to 4 cycles of treatment were associated with shorter progression-free survival and overall survival, respectively. CONCLUSIONS The intermittent erlotinib schedule was associated with a higher response rate, although this is not statistically significant. Serum TGFa and AREG levels have prognostic significance in erlotinib-treated patients with colorectal cancer, and further studies are warranted. ; Department of Health Technology and Informatics