OBJECTIVE: To investigate the Optimal Screening for Prediction of Referral and Outcome Yellow Flag (OSPRO-YF) tool for longitudinal monitoring of pain associated distress with the goal of improving prediction of 50% reduction in pain intensity and disability outcomes. DESIGN: Cohort study with 12-month follow-up after initial care episode. SETTING: Ambulatory care, participants seeking care from outpatient physical therapy clinics. PARTICIPANTS: Participants (N=440) were seeking care for primary complaint of neck, low back, knee, or shoulder pain. This secondary analysis included 440 subjects (62.5% female; mean age, 45.1±17y) at baseline with n=279 (63.4%) providing follow-up data at 12 months. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: A 50% reduction (baseline to 12-mo follow-up) in pain intensity and self-reported disability. RESULTS: Trends for prediction accuracy were similar for all versions of the OSPRO-YF. For predicting 50% reduction in pain intensity, model fit met the statistical criterion for improvement (P<.05) with each additional time point added from baseline. Model discrimination improved statistically when the 6-month to 12-month change was added to the model (area under the curve=0.849, P=.003). For predicting 50% reduction in disability, there was no evidence of improvement in model fit or discrimination from baseline with the addition of 4-week, 6-month, or 12-month changes (P>.05). CONCLUSIONS: These results suggested that longitudinal monitoring improved prediction accuracy for reduction in pain intensity but not for disability reduction. Differences in OSPRO-YF item sets (10 vs 17 items) or scoring methods (simple summary score vs yellow flag count) did not affect predictive accuracy for pain intensity, providing flexibility for implementing this tool in practice settings.