Aims: Defining the most appropriate definition of mesorectal fascia involvement (MRF+) by reviewing literature and using new inclusion criteria to re-evaluate the effectiveness of MRI in the assessment of MRF+ for rectal cancer. Methods: PubMed, Medline, Embase, and the Cochrane Library databases were electronically searched from 1999 to 2012. The bivariate random effects model was used to estimate the pooled outcomes of each subgroup. The definition of MRF+ in MRI and the influence of neoadjuvant chemoradiotherapy (neo-ChRT) were especially discussed. Results: Fourteen studies involving 1,600 patients were included. Different definitions of MRF+ (≤1, ≤2 and ≤5 mm) in MRI exhibited different pooled sensitivity (76, 79 and 92%), specificity (88, 66 and 48%) and diagnostic odds ratio (DOR) (22.4, 6.6 and 16.0). The definition of MRF+ at ≤1 mm showed the highest DOR. The specificity (88 vs. 93%, p = 0.026) and DOR (15.5 vs. 39.0, p = 0.001) were lower in patients who underwent neo-ChRT than those who did not while using ≤1 mm as the definition of MRF+. However, the sensitivity showed no significant difference (67 vs. 74%, p = 0.129). Conclusions: MRI is valuable for the assessment of MRF. The most appropriate definition of MRF+ in MRI is ≤1 mm. The effectiveness is higher in patients who did not undergo neo-ChRT.