BackgroundThe claw sign is advocated as a discriminant of renal versus non-renal origin of tumours. The accuracy of the claw sign on magnetic resonance imaging (MRI) is unknown and is potentially hindered by the inferior spatial resolution and the larger tumour sizes at presentation in developing countries.ObjectivesTo define and evaluate the claw sign in differentiating renal from non-renal retroperitoneal masses in children undergoing MRI.MethodsA definition of the claw sign was proposed. Magnetic resonance imaging studies, clinical and laboratory records of 53 children were reviewed to test the diagnostic accuracy, inter- and intra-observer reliability. Three tumour–mass interface characteristics, inherent to the claw sign, were tested: (1) a smooth tapering kidney edge blending continuously with the tumour, (2) absence of infolding of the kidney and (3) an obtuse superficial angle.ResultsThe sensitivity, specificity, negative predictive value and positive predictive values of the claw sign were 97%, 74%, 83% and 94%. The Cohen’s kappa values for intra-rater reliability were 0.72 (95% confidence interval 0.54–0.86) for the first reader and 0.83 (0.66–1.00) for the second reader. The Cohen’s kappa values for inter-rater reliability were 0.67 (0.50–0.85) and 0.65 (0.44–0.86) for the second reading respectively (p < 0.0001).ConclusionThe three tumour–mass interface characteristics investigated are all important characteristics of the claw sign. Intra- and inter-rater reliability is moderate to strong for all characteristics and overall impression of the claw sign. The claw sign is therefore sensitive in the accurate placement of an intra-renal mass but lacks specificity.