The objective of this study was to describe the health care utilization and prospective predictors of high-cost primary-care back pain patients. In the primary-care clinics of a large, staff model health maintenance organization in western Washington State, 1059 subjects were selected from consecutive patients presenting for back pain. The design was a 1-year prospective cohort study. Patients' were interviewed 1 month after an index primary-care back pain visit. Costs (back pain and total) and utilization (back pain primary-care follow-up visits, back pain specialty visits, back pain hospitalizations, back pain radiologic procedures, and pain medicine fills) were tracked over the next 11 months. Predictors assessed at 1 month were back pain diagnosis (disc disorder/sciatica, arthritis, vs. other), chronic pain grade (measure of pain intensity and related dysfunction), pain persistence (days with pain in prior 6 months), depressive symptomatology, and back pain-related disability compensation (ever/never). For the sample, 21% of patients with back pain costs > or = $600 (high back pain costs) accounted for 66% of back pain costs, 42% of total costs, 55% of primary-care follow-up visits for back pain, 91% of back pain specialty visits, 100% of back pain hospitalizations, 51% of back pain radiologic procedures, and 52% of pain medicine fills. The 21% with total costs > or = $2700 (high total costs) accounted for 67.7% of total costs, 52% of back pain costs, 29% of primary-care follow-up visits for back pain, 66% of back pain specialty visits, 100% of back pain hospitalizations, 39% of back pain radiologic procedures, and 42% of pain medicine fills. Multivariable logistic regression analyses indicated that increasing chronic pain grade, more persistent pain, and disc disorder/sciatica were strong independent predictors of high total and high back pain costs. Increasing depressive symptoms significantly predicted high total but not high back pain costs. Back pain disability compensation predicted high back pain but not high total costs. A minority of primary-care back pain patients accounted for a majority of health-care costs. Patients with high back pain costs accounted for more back pain-related health-care utilization than did patients with high total costs. Factors predicting subsequent high costs suggest behavioral interventions targeting dysfunction, pain persistence, and depression may reduce health-care utilization and prevent accumulation of high health-care costs among primary-care back pain patients.