To analyze the reliability and validity of magnetic resonance imaging (MRI) for the detection of anterior talofibular ligament (ATFL) injuries in chronic lateral ankle instability by comparing its findings with arthroscopic findings. This diagnostic study enrolled patients who underwent MRI followed by subsequent arthroscopy for their various ankle disorders between April 2012 and February 2013. Two radiologists independently assessed the ATFL on MRI, and the results of their MRI assessments were then compared with the arthroscopic findings, which were used as the standard of reference. On arthroscopy, 55 ATFL injuries were identified in 79 patients. The interobserver reliability of detecting ATFL injuries with MRI was excellent (intraclass correlation coefficient, 0.915). MRI, as interpreted by readers A and B, showed a sensitivity of 83.6% and 76.4%, respectively; specificity of 91.7% and 83.3%, respectively; negative predictive value of 71.0% and 60.6%, respectively; positive predictive value of 95.8% and 91.3%, respectively; and accuracy of 86.1% and 78.5%, respectively. According to the location of the ATFL injury, the sensitivity of MRI for readers A and B was 72.7% and 63.6%, respectively, at the fibular attachment site; 80.0% and 66.7%, respectively, at the talar attachment site; and 100% at the midsubstance and multiple sites. All false-negative diagnoses of ATFL injuries were observed at the fibular or talar attachment site (9 cases for reader A and 13 cases for reader B). This study showed that MRI has excellent interobserver reliability (intraclass correlation coefficient, 0.915) for detecting ATFL injuries in patients in whom there is a clinical suspicion of chronic lateral ankle instability. The sensitivity and positive predictive value of MRI in the diagnosis of ATFL injuries were very high, whereas the sensitivity and negative predictive value of MRI were relatively low. According to the location of the ATFL injury, the sensitivities of MRI for the detection of ATFL injuries at the fibular or talar attachment site were lower than those at the midsubstance or multiple sites. In addition, all false-negative diagnoses of ATFL injuries were observed at the fibular or talar attachment site. Level III, diagnostic study of nonconsecutive patients (without consistently applied reference gold standard).