Some patients with low back pain recover quickly while others continue to experience pain beyond 3 months. The primary aim of this study was to develop a simple prediction rule to help clinicians identify patients with acute low back pain likely to recover at different rates. The secondary aim was to compare a clinician's prognosis judgement to the prediction rule. The study sample included 239 patients with acute low back pain who participated in a randomised controlled trial. The primary outcome was days to recovery from pain. Potential prognostic factors were initially tested for univariate association with recovery using Cox regression (p < 0.1). Continuous prognostic factors with a significant association were dichotomised using a median split. Significantly associated prognostic factors were then included in a multivariate forward stepwise Cox regression. We then separated participants into strata according to the number of predictors in the final model for which they were positive. Our results suggest that a clinical prediction rule using three simple prognostic factors was able to differentiate between patients who recover quickly and those who recover slowly. Patients with lower than average initial pain intensity, shorter duration of symptoms and fewer previous episodes recovered more quickly (HR = 3.5, 95% CI, 1.8-7.0) than patients without these characteristics. Therapists were able to predict patients likely to recover at different rates, (HR = 1.6, 95% CI, 1.2-2.1), however, they did not perform as well as the clinical prediction rule. The rule requires validation in a different sample of patients.