OBJECTIVE: To evaluate the course of symptoms and occupational outcomes in low back pain patients two years after conservative inhospital therapy. METHODS: Retrospective medical chart study of all patients admitted to a rheumatology department in Geneva in 1993, and telephone interview two years after the admission. RESULTS: Eighty of the 99 patients (81%) identified by the chart review were interviewed by telephone. The course of symptoms (evaluated based on subjective assessments and on the number of subsequent admissions) was similar in the patients who had acute (n = 33) and chronic (n = 47) low back pain at the index admission. Patients who had not returned to work were significantly more likely to report persistent pain (P < 0.001). As many as 61% of patients had not resumed work. Sick leave duration and symptom duration at admission were the factors that had the largest effects on prognosis. Over 90% of patients who were put on sick leave for more than four weeks did not return to work. Permanent cessation of work was recorded in 82% of manual laborers versus 27% of patients in sedentary jobs (P < 0.001). CONCLUSION: Our results are ascribable, at least in part, to the fact that our population included large numbers of patients with persistent pain and of blue collar workers with physically-challenging jobs. Sick leave duration and occupation were the best predictors of the course of symptoms and work status.