A role of maintenance chemotherapy (mCT) in patients (pts) with metastatic colorectal cancer (mCRC) is still controversial. The purpose of this retrospective study was to investigate the toxicity and efficacy of mCT in pts with mCRC. There were 97/291 (33 %) pts with mCRC completed 18-20 weeks of first-line CT from 2007 to 2013 in our center. Then, pts who had no disease progression were non-randomly allocated to mCT with capecitabine ± bevacizumab (n = 35) or surveillance (n = 62). PFS was used as a primary endpoint and was calculated from the date of completion of first-line CT. Multivariate Cox stepwise regression analysis was performed to determine independent prognostic factors. Median follow-up time was 15 (range 5-60) months. Median PFS and OS were higher in pts with mCT: 7 versus 3 months (HR 0.5, 95 %CI 0.28-0.82, p = 0.007) and 29 vs 16 months (HR 0.6, 95 %CI 0.3-1.1, 0.04-Gehan-Breslow-Wilcoxon test). Following independent negative prognostic factors was significant on multivariate analysis: CEA level >2.5 ng/ml before start of first-line CT (p = 0.02), liver metastases (p = 0.03) and number of metastatic zones >2 (p = 0.008). MCT had an independent positive impact on PFS (HR 0.5, p = 0.003). MCT prolonged PFS in pts with at least one negative prognostic factors (7 vs. 3 months, p = 0.001, HR 0.38, 95 % CI 0.22-0.68). The mCT was most beneficial in pts with negative prognostic factors: CEA level >2.5 ng/ml before start of first-line CT and/or liver metastases and/or number of metastatic zones >2.