Objective: To determine the in vivo ability to predict uric acid stone composition by Hounsfield units (HU) with the addition of urine parameters. Methods: We reviewed all consecutive stones sent for analysis during a 4-year period from our institution for patients with an in-house computed tomography (CT) scan within the prior 6 months and urinalysis within the prior week. CT scans were independently reviewed by a radiologist blinded to stone composition. Results: Of the 507 patients with stones sent for analysis, 235 met the criteria for inclusion. Analysis showed 212 stones were predominantly calcium-based, and 22 were predominantly uric acid in composition. There was a significant difference between calcium stones and uric acid stones in mean HU (890 � 20 vs 484 � 44; P <.01) and urine ph (6.4 � 0.8 vs 5.1 � 0.2; p><.01). receiver operating characteristic curve evaluation gave optimal predictive values of hu ?494 (rounded to 500) and ph of ?5.5 to predict uric acid stones. the combination of hu and ph criteria resulted in a sensitivity of 86% and a specificity of 98%, with a positive predictive value of 80%, which increased to 90% if we limited to stones>4 mm. Conclusion: Uric acid stones show a significant difference in HU and urine pH from calcium stones, and the use of both criteria is superior to either separately. For a stone >4 mm, a HU ?500 and pH ?5.5 has a positive predictive value of 90% for uric acid composition. Our prediction model gives a straightforward tool that can be easily measured to predict a uric acid stone.