BACKGROUND Multiple somatic symptoms are common and may cause prolonged sickness absence (SA) and unsuccessful return to work (RTW). OBJECTIVE To compare three instruments and their predictive and discriminative abilities regarding RTW. METHODS A longitudinal cohort study of participants recruited from two municipal job centres, with at least 8 weeks of SA. The instruments used were the Symptom Check List of somatic distress (SCL-SOM) (score 0-48 points), the Bodily Distress Syndrome Questionnaire (BDSQ) (0-120 points) and the one-item self-rated health (SRH) (1-5 points). The instruments' predictive value was explored in a time-to-event analysis. Different cut-points were analysed to find the highest number of correctly classified RTW cases, identified in a register on public transfer payments. RESULTS The study involved 305 subjects. The adjusted relative risk regarding prediction of RTW was 0.89 [95% confidence interval (CI) 0.83-0.95], 0.89 (95% CI 0.83-0.95) and 0.78 (95% CI 0.70-0.86) per 5-, 10- and 1-point increase in the SCL-SOM, BDSQ and SRH, respectively. After mutual adjustment for the three instruments, only the prediction of RTW from SRH remained statistically significant 0.81 (95% CI 0.72-0.92). The highest sensitivity (86%) was found by SRH at the cut-point ≤5, at which 62% were correctly classified. CONCLUSIONS All three instruments predicted RTW, but only SRH remained a significant predictor after adjustment for the SCL-SOM and BDSQ. The SRH provides an efficient alternative to more time-consuming instruments such as SCL-SOM or BDSQ for estimating the chances of RTW among sickness absentees.