OBJECTIVE: The purpose of this study was to evaluate whether the predictive validity of the Dutch version of the STarT Back Tool (SBT) can be improved by (1) using other cut-off values, (2) changing the items, or (3) adding prognostic factors to the SBT. DESIGN: This was a secondary analysis of a prospective cohort study (PRINS study: Prevalence of Risk groups in Neck- and back pain patients according to the STarT back screening tool) in patients with low back or neck pain. METHODS: The predictive validity was calculated with a relative risk ratio and a Spearman correlation. The new cut-off values were calculated with receiver operating characteristic curves. Replacing items of the SBT and adding new items were assessed with logistic regression analyses. RESULTS: A total of 150 patients were included; 51% were categorized as having low risk, 39% as moderate risk, and 11% as high risk. Changing the cut-off total score to ≤2 and the subscore to ≥5 led to an improvement of the Spearman correlation and RR. Adding the item "duration of the complaints" improved the RR for moderate risk (3.6) (95% CI = 1.6-7.9) and for high risk (9.0) (95% CI = 4.2-19.1) compared with low risk. The new Spearman correlation was improved to rs = 0.37. CONCLUSION: The predictive validity was improved by adding the item "duration of the complaints" and changing the cut-off values.