BACKGROUND: Studies have shown that positive recovery expectations are associated with positive health outcomes in patients with chronic low back pain (CLBP) such as return to work (RTW) and the time to RTW. AIM: To compare the predictive value for RTW in CLBP using different subsets of the Orebro Musculoskeletal Pain Screening Questionnaire (OMPSQ). DESIGN: Longitudinal cohort study. SETTING: Rehabilitation center. POPULATION: 98 inpatients with CLBP (> 3 months). METHODS: The OMPSQ at baseline was used to predict RTW three months after discharge from the rehabilitation clinic. The area under the ROC-curve was calculated based on a logistic regression model. Cox-regression was used to analyze time to RTW with C statistics for the original full (25-items) version of the OMPSQ, the 10-item version as well as for the two items about self-expected RTW and self-expected recovery. RESULTS: The area under the curve (AUC) for the overall score of the full version OMPSQ was 0.82 (95% CI 0.73 to 0.90), the AUC for the short version was 0.79 (95% CI 0.70 to 0.88), the AUC for the item about self-expected recovery (#15) was 0.67 (95% CI 0.57 to 0.78), and the AUC for the item about self-expected RTW (#16) was 0.76 (95% CI 0.66 to 0.85). Harrell's C for the full version was 0.74 (95% CI 0.66 to 0.81), for the short version the C was 0.71 (95% CI 0.64 to 0.79), for the item #15 the C was 0.62 (95% CI 0.53 to 0.72), and for the item #16 the C was 0.71 (95% CI 0.64 to 0.78). CONCLUSIONS: Two items about expectations from the OMPSQ showed similar predictive value for RTW compared to the short and full original versions, and could be used as first screening questions. CLINICAL REHABILITATION IMPACT: Clinicians may make an informed choice whether they use the full or the short version of the OMPSQ for screening of psychosocial problems, or whether they use the two single items about expectations. Knowledge about patient's expectations provides a base for discussion between health professionals and the patient.