Background: A large proportion of patients with colorectal cancer liver metastases (CRCLM) not amenable to curative liver resection will progress on systemic therapy. Intra-arterial therapies (IAT) including conventional transarterial chemoembolization (cTACE), drug eluting beads (DEB-TACE) and yttrium-90 radioembolization (Y-90) are indicated to prolong survival and palliate symptoms. The purpose of this systematic review and meta-analysis is to compare the survival benefit and radiologic response of three intra-arterial therapies in patients with chemorefractory and unresectable CRCLM. Methods: A systematic search for eligible references in the Cochrane Library and the EMBASE, MEDLINE and TRIP databases from January 2000 to November 2016 was performed in accordance with PRISMA guidelines. Methodological quality of included studies was assessed using the MINORS scale. One-year overall survival rates and RECIST responder rates were pooled using inverse-variance weighted random-effects models. Overall survival outcomes were collected according to transformed pooled median survivals from first IAT with a subgroup analysis of patients with extrahepatic disease. Results: Twenty-three prospective studies were included and analyzed: 5 cTACE (n = 746), 5 DEB-TACE (n = 222) and 13 Y-90 (n = 615). All but five were clinical trials. Eleven of 13 Y-90 studies were industry funded. Pooled RECIST response rates with 95% confidence intervals (CI) were: cTACE 23% (9.7, 36), DEB-TACE 36% (0, 73) and Y-90 23% (11, 34). The pooled 1-year survival rates with CI were: cTACE, 70% (49, 87), DEB-TACE, 80% (74, 86) and Y-90, 41% (28, 54). Transformed pooled median survivals from first IAT and ranges for cTACE, DEB-TACE and Y-90 were 16 months (9.0–23), 16 months (7.3–25) and 12 months (7.0–15), respectively. Significant heterogeneity in inclusion criteria and reporting of confounders, including previous therapy, tumor burden and post-IAT therapy, precluded statistical comparisons between the three therapies. Conclusion: Methodological and statistical heterogeneity precluded consensus on the optimal treatment strategy. Given the common use and significant cost of radioembolization in this setting, a more robust prospective comparative trial is warranted.