Background: Palliative capecitabine (X) monotherapy for advanced colorectal colorectal cancer (aCRC) is generally well tolerated by elderly or frail pts. Epidermal growth factor receptor (EGFR) monoclonal antibodies improve efficacy when added to combination chemotherapy for mCRC. Erlotinib (E), an oral EGFR tyrosine kinase inhibitor (TKI) may add benefit when added to X. We conducted a randomized phase II trial to investigate the novel “all-oral” combination of X and E in aCRC. Methods: Pts with untreated aCRC who were either deemed unfit for, or chose against, combination chemotherapy were randomized to X (1000 mg/m² PO BID x 14 days) alone or in combination with E (150 mg PO OD) on a 3-week schedule. Primary endpoint was time to disease progression (TTP); secondary endpoints included: objective response rate (ORR), overall survival (OS), and safety. Tumours were designated as left-sided if there were distal to the transverse colon. KRAS status was retrospectively analyzed for 72 of 82 pts. Results: From 2004 to 2008, 82 pts were randomized to X alone (40 pts) or XE (42 pts). TTP was not different between X and XE (7.9 m vs. 9.2 m; p = 0.890), however, KRAS subgroup analysis revealed pts with KRAS mutations did significantly worse when treated with XE compared to X alone (1.9m vs 7.4m; HR = 2.63; P = 0.038). Pts with KRAS wild-type (WT) treated with XE had an improved TTP compared to those treated with X (11.7m vs. 8.4m, HR = 0.73; Wilcoxon P = 0.061). KRAS-WT pts treated with XE with left-sided disease had an improved OS compared to pts with right-sided disease (16m vs. 12.1m, not significant). KRAS-WT pts with left-sided primaries treated with XE had a non-significant improvement in TTP compared to pts treated with X alone (11.7m vs 8.4m). XE was well tolerated but had significantly higher rates of diarrhea and acneform skin rash. Conclusions: The addition of E to X is generally well tolerated but associated with additional toxicities. E may benefit pts with KRAS-wild-type CRC, specifically those with left-sided primary tumours, and likely harms those with KRAS-mutated CRC. Further study of oral EGFR-TKIs in left-sided KRAS-wild-type aCRC is warranted.