OBJECTIVES: To investigate the efficacy and safety of second-line chemotherapy in mCRC and to determine the feasibility of relative efficacy between treatments using indirect or mixed treatment comparisons. METHODS: An SLR on randomized trials published from 1992–2012 was performed. EMBASE, PubMed, CENTRAL databases were consulted. Manual searches (including ASCO and ESMO abstracts) were also conducted. Eligible studies evaluated second-line clinical efficacy and safety endpoints in mCRC.  RESULTS: Twenty-seven trials with patients enrolled ranging from n=33–1,226 were identified. Patients were markedly diverse in terms of ECOG, EGFR, or KRAS status, and prior treatment exposures: oxaliplatin (k=9), irinotecan +/- 5-FU (k=5), bevacizumab (k=7), and 5-FU (k=22). On-study treatments and comparators were also diverse: FOLFIRI alone (t=6) or with panitumumab (t=2) or aflibercept (t=1); FOLFOX4 alone (t=6) or with bevacizumab (t=1); bevacizumab alone (t=1) or with chemotherapy (t=2); irinotecan alone (t=19) or with FOLFOX4 (t=2) or cetuximab (t=1); and miscellaneous other regimens (t=19). Study methodologies varied in analysis populations (ITT [k=24] vs. per protocol or not reported [k=3]), primary efficacy outcomes (OS, PFS, and/or tumor response, as variably measured), follow-up durations, and timing of outcomes measurements. Preliminary network diagrams to assess feasibility of meta-analyses for ITC revealed only a single study for each treatment comparison of interest. Assumptions of comparability of apparently diverse patients, past and current treatment regimens, and comparator arms would be necessary to pool treatment groups for meta-analysis of treatment effectiveness across studies and hence, is not feasible. Given the lack of SOC treatments used in many trials, introducing such assumptions will compromise the evidence-based criteria for quantitative clinical data syntheses for indirect or mixed treatment comparisons. CONCLUSIONS: Current standards around second-line treatment of mCRC have evolved over time. The substantial clinical, methodological, and statistical heterogeneity in the available data prevents evidence-based quantitative comparisons of treatment outcomes at this time.