To detect painful vertebral fractures (VFs) in back pain populations at risk of osteoporosis, we designed a physical examination test (the Back Pain‐Inducing Test [BPIT]) that included three movements:​ lying supine, rolling over, and sitting up. If back pain is induced during any of these movements, the result is defined as positive, thereby establishing a presumptive diagnosis of painful VFs. Pain severity is quantified using a self‐reported numerical rating scale (NRS). The presence or absence of painful VFs is verified by whole‐spine magnetic resonance imaging (MRI), the gold standard for final diagnosis. According to the standards for reporting diagnostic accuracy, a real‐world, prospective, and observational study was performed on 510 back pain patients (enrolled from a single institute) at risk of osteoporosis. The sensitivity, specificity, and accuracy of the BPIT for identifying painful VFs were 99.1% (95% CI, 97.5% to 99.8%), 67.9% (95% CI, 60.4% to 74.5%), and 89.0%, respectively. The positive and negative predictive values were 86.6% (95% CI, 82.9% to 89.6%) and 97.4% (95% CI, 92.6% to 99.3%), respectively. Cutoff NRS scores for lying supine, rolling over, and sitting up were 3, 0, and 2, respectively. The corresponding area under the receiver operating characteristic curves (AUROCs) of each movement was 0.898 (95% CI, 0.868 to 0.922), 0.884 (95% CI, 0.854 to 0.911), and 0.910 (95% CI, 0.882 to 0.933), respectively. Although the high prevalence of VFs in the enrolled cohort partially limits the external validity of the predictive value in the general population, we conclude that the BPIT is potentially effective for detecting painful VFs in back pain populations at risk of osteoporosis. This test may be used as a stratification tool in decision‐making on subsequent imaging procedures:​ a negative BPIT rules out painful VFs and indicates that an MRI should be spared, whereas a positive BPIT means that an MRI is necessary and is likely to identify painful VFs.