Background: Fruquintinib with PD-1 inhibitors (FP) and TAS-102 with bevacizumab (TB) are two common therapies for patients with previous-treated metastatic colorectal cancer (mCRC). However, it’s still not clear that which therapy can bring better prognosis. Methods: This is a retrospective cohort study conducted in Hunan Cancer Hospital. Patients (pts) with mCRC who had received at least the 2^nd line treatment were eligible. Propensity score (PS) would be calculated to balance the baseline characteristics of two arms. Overall survival (OS) was set as the primary endpoint. Results: From July 2019 to October 2022, 106 eligible pts in total were enrolled. According to the treatment received, 72 and 34 pts were respectively allocated into FP cohort and TB cohort. With a global median follow-up of 14, the crude OS was 19.4 (95% CI: 17.9-NA) mo in FP cohort vs. 11.6 (10.0-17.2) mo in TB cohort. The HR of FP was 0.384 (95% CI: 0.192-0.769, TB as reference). Multivariate Cox regression showed that the adjusted HR was 0.323 (95% CI: 0.149-0.704), which was adjusted with sex, age greater than 65 yr, ECOG-PS, resection of primary lesion, radiotherapy history, location of lesion (left or right), metastases (liver, lung, and bone), and current line (3 or later). With PS, we performed three statistical methods, namely inverse probability weighting, PS matching (the sample size was 49 vs. 29 after matching), and additional adjustment for PS with multivariate cox regression. The HRs of FP were 0.437 (95% CI: 0.200-0.953), 0.446 (95% CI: 0.201-0.990), and 0.339 (95% CI: 0.153-0.748), respectively. Subgroup analysis indicated that FP demonstrated lower HR in subgroup of male (0.296, 95% CI: 0.117- 0.753), right-sided lesion (0.177, 95% CI: 0.043- 0.736), liver metastasis (0.291, 95% CI: 0.112- 0.756), and current line greater than 3 (0.303, 95% CI: 0.113- 0.812). Additionally in the raw set, the ORR was 16.67% vs. 8.82% (OR_FP=2.067, 95% CI: 0.602-9.559), and the DCR was 93.06% vs. 73.53% (OR_FP=4.824, 95% CI: 1.518-17.030). Conclusions: FP demonstrated robustly lower HRs for OS in both crude and PS analysis. Likewise, FP showed better OS and DCR in pts with late-line mCRC than TB, which suggested FP might be a better therapy in late-line mCRC.