Thirty-one cases of flexor hallucis longus injuries in 26 patients were treated over a 16-year period (1977-1993). Groups were divided into dance-related injuries (group I) and other causes (group II). The two groups were compared with regard to age, activity, duration of symptoms, operative findings, histopathology, and postoperative time to resumption of full activities. Twenty-seven cases required surgery for unsuccessful nonoperative treatment. In group I, 71% of patients had a partial longitudinal tear of the flexor hallucis longus compared with 30% in group II. Another common finding was isolated tenosynovitis (21% in group I and 53% in group II). Eight cases had magnetic resonance imaging (MRI) evaluations before surgery. Clinical correlation was found to be an important factor in interpreting the MRI. Dancers tended to have symptoms for a longer period of time before seeking treatment than did nondancers. Follow-up was 19.2 months for dancers and 25 months for nondancers. Surgical correction of tenosynovitis, pseudocyst, and tendon tear yielded good or excellent results in 14 of 15 dancers and 9 of 11 nondancers. Surgical treatment of tendon tears and other pathologic tendon conditions gave consistently good results in patients with refractory flexor hallucis longus disease.