Abstract Revascularization of chronic total occlusion (CTO) is still debated regarding its indications and therapeutic benefits. Guidelines recommend patient selection based on ischemia detection and viability assessment. We aimed to investigate the relationship between the quality of coronary collaterals (CC), graded by Rentrop classification, and myocardial viability assessed by cardiovascular magnetic resonance (CMR). Unselected 100 consecutive patients with a single CTO were prospectively enrolled. CC of Rentrop grade two or three were considered as well-developed. Analyzing late gadolinium enhancement (LGE) images, CTO territories with mean segmental transmural scar extent < 50% were considered viable. Of the 100 patients (70 male, mean age 58.0 ± 6 years), 73 patients (73%) had angiographically visible CC. Based on LGE, patients were classified into viable (n = 50) and non-viable (n = 50) groups. Significant differences between both groups existed regarding frequency of diabetes mellitus (p = 0.044), frequency of congestive heart failure (p = 0.032), presence of pathological Q in CTO territory (p = 0.039); and presence of well-developed CC (p < 0.001). Binary logistic regression and receiver operating characteristic curve showed that presence of well-developed CC could independently (OR 9.4, 95% CI: 2.6–33.6, p < 0.001) predict myocardial viability with a sensitivity and a specificity of 72% and 74%, respectively (AUC: 0.796, 95% CI: 0.708—0.884, P < 0.001). The presence of well-developed CC could independently predict with high accuracy myocardial viability assessed by LGE in territories subtended by CTO vessels. Therefore, search for viable myocardium using different imaging modalities, e.g. CMR, may be recommended in CTO patients with well-developed CC.