Background Prevention of chronic pain relies on accurate detection of at-risk patients. Screening tools have been validated mainly in (sub) acute spinal pain and the need of more generic tools is high. We assessed the validity of the French version of the short Örebro Musculoskeletal Pain Screening Questionnaire (ÖMPSQ) in patients with a large range of pain duration and localization. Methods First, we re-analyzed data from a 6-month longitudinal study of 73 patients with (sub) acute spinal pain consulting in secondary line settings. Secondly, we performed a new 12-month longitudinal study of 542 primary care patients with (sub) acute and chronic pain in different localizations (spinal, limbs, “non-musculoskeletal”). The area under the receiver operating characteristic curve and cutoff scores were computed and compared for different subpopulations and ÖMPSQ subscores. Results Data from patients suffering from (sub) acute and chronic spinal pain consulting in both primary and secondary care settings confirmed the validity of the short French ÖMPSQ version and its subsets. In the primary care cohort, the performance of the questionnaire and its psychosocial subscore was variable but at least “fair” in most populations ((sub) acute and chronic, spinal and limb pain). Cutoff scores showed quite large variability depending on the outcome and the subpopulation considered. Conclusions These results confirm the usefulness of the short French ÖMPSQ for prediction of the evolution of (sub) acute and chronic patients with spinal and limb pain, whatever its duration. However, increasing population heterogeneity results in slightly worse predictive performance and largely variable cutoff scores. Consequently, it might be difficult to choose universal cutoff scores and other criteria, such as patients’ values and the available resources for patient management, should be taken into account.