PURPOSE: The purpose of this study was to show that magnetic resonance imaging (MRI) can accurately and noninvasively evaluate the articular cartilage overlying osteochondral lesions of the talus, using arthroscopy as the standard, and provide information about lesion stability. A classification system applicable to both MRI and arthroscopy is proposed. TYPE OF STUDY: Retrospective review with MRI and arthroscopy correlation. METHODS: A retrospective analysis involving 26 men and 28 women who underwent MRI of the ankle and in whom arthroscopy had been performed. The talar dome cartilage was evaluated using the following grading system: 0, normal cartilage; 1, abnormal signal but intact; 2, fibrillation or fissures not extending to bone; 3, flap present or bone exposed; 4, loose undisplaced fragment; 5, displaced fragment. There was a history of trauma in 80% of cases. RESULTS: At surgery, 14 ankle joints had intact cartilage and 40 showed osteochondral lesions. MRI correctly identified all 40 osteochondral lesions and all 14 normal ankles. MRI correctly graded 33 of 40 (83%) of the osteochondral lesions using the described 5-point scale. Of the remaining 7 lesions, all were identified within 1 grade. Collapsing all grades into disease-negative status (grades 0 and 1) and disease-positive status (grades 2, 3, 4, and 5) yielded sensitivity of 95%, specificity of 100%, negative predictive value of 88%, and positive predictive value of 100%. CONCLUSIONS: This study shows that MRI can be used to evaluate the articular cartilage covering osteochondral lesions of the talus with a high degree of accuracy and thus may aid in identifying patients who would benefit from definitive arthroscopic treatment.