OBJECTIVE There is no set of prospectively validated criteria to identify the emergency department (ED) patients with renal colic who are most likely to eventually have to undergo an intervention. This study prospectively assessed predictors of intervention in this patient population. METHODS This prospective cohort study included adult patients with renal colic who presented to 2 tertiary care hospital EDs. Patients had an 18-variable data form completed by an emergency physician and a radiological study to confirm urolithiasis. After discharge, patients were followed at 1 and 4 weeks to assess for intervention. The outcome criteria included the patient having had at least 1 of the following procedures performed: extracorporeal shockwave lithotripsy (ESWL), ureteroscopy, percutaneous nephrostomy or open surgery. Data were analyzed using appropriate univariate techniques, and those variables associated with intervention were combined using logistic regression analysis. RESULTS Over an 8-month period, 245 patients with confirmed urolithiasis were followed; 20% (95% confidence interval [CI] 15%-25%) eventually had a procedure to remove their calculi. Three variables were significantly correlated with having a procedure: i) size of calculus >or= 6 mm (odds ratio [OR] 10.7, 95% CI 4.6-24.8), ii) location of calculus above mid-ureter (OR 6.9, 95% CI 3.0-15.9), and iii) Visual Analogue Scale score for pain at discharge from the ED >or= 2 cm (OR 2.6, 95% CI 1.0-6.8). The area under receiver operating characteristic curve was 0.77 (95% C I 0.70-0.84) (p < 0.001). If all variables were present there was a 90% probability of the patient having an intervention performed within 4 weeks of discharge from the ED. Conversely, if none of the variables were present there was only a 4% probability of an intervention. Overall, the model had a sensitivity of 92% (95% CI 89%-96%) and a specificity of 63% (95% CI 57%-69%). CONCLUSIONS This study has identified variables that could potentially be used to identify those renal colic patients who require an intervention after ED evaluation. Future studies will prospectively validate this model.