Standard chemotherapy for one year was proposed for the treatment of tuberculous patients in French Polynesia between 1971 and 1978, and for nine months with the addition of pyrazinamide from 1979 to 1984. After treatment, the patients were followed up systematically for a period of 5 years. The aim of this study was to evaluate the value of the follow up measures to detect relapses. Of 1,065 patients treated between 1971 and 1984 at the centre for the control of tuberculosis 35 (3%) presented with a relapse, of whom 30 were confirmed either by bacteriological examination or histological examination. The level of relapse seen was not linked to age, sex, nor type of treatment received (one year or nine months). However, the level of relapse in non compliant patients to the initial treatment (7%) was significantly greater than those observed in patients who were totally compliant (2%) (p less than 0.0001). Of the 35 relapses 9 (26%) were discovered during routine follow up visits, of whom 8 were detected during the first year. The median delay separating apparent recovery from relapse was 8 months. 28 patients (80%) presented with symptoms suggestive of a relapse. Over 5 years 5773 consultations were carried out, which meant 642 consultations to detect one relapse. The authors suggest that in French Polynesia a systematic follow up limited to one year after apparent cure should be adequate.