The identification of novel survival predictors may help to improve the appropriate management of colorectal cancer (CRC). In the present study, two gene sets associated with irinotecan or oxaliplatin resistance in CRC cell lines were first identified and subsequently applied to the clinical CRC microarray dataset GSE14333. Subsequently, a 60-gene irinotecan resistance-associated signature and a 13-gene oxaliplatin resistance-associated signature were established, which were able to classify CRC patients into high- and low-risk subgroups with varied clinical outcomes [irinotecan-resistance gene signature: hazard ratio (HR)=0.4607, 95% confidence interval (CI)=0.3369–0.6300, P<0.0001; oxaliplatin-resistance gene signature: HR=0.6119, 95% CI=0.4547–0.8233, P=0.0008]. The performance of these two gene expression signatures in predicting outcome risk were also validated in two other independent CRC gene expression microarray datasets, GSE17536 (irinotecan-resistance gene signature: HR=0.5318, 95% CI=0.3359–0.8419, P=0.0079; oxaliplatin-resistance gene signature: HR=0.5383, 95% CI=0.3400–0.8521, P=0.0114) and GSE17537 (irinotecan-resistance gene signature: HR=0.2827, 95% CI=0.1173–0.6813, P=0.0088; oxaliplatin-resistance gene signature: HR=0.2378, 95% CI=0.09773–0.5784, P=0.0023). Furthermore, the combination of these two gene classifiers demonstrated a superior performance in CRC prognosis prediction than either used individually. Therefore, this study proposed novel gene classifier models for CRC prognosis prediction, which may be potentially useful to inform treatment decisions for patients with CRC in clinical settings.