Objective: To assess the validity of bone age assessment by ultrasonography (US). Study design: Wrist US was performed on children (n = 100) undergoing radiographic bone age and compared with bone age estimation by a radiologist in the clinic and by endocrinologists under blinded conditions with Greulich and Pyle (GP) and Tanner and Whitehouse (TW3) methods. Results: The strongest correlation (r2) was seen in the radiographic bone age assessment between the 2 endocrinologists using the GP method (96.7%). The poorest correlation was seen when comparing radiographic methods to US of either wrist (74.6% to 82.6%). When bone age correlations were divided into normal, delayed or advanced, the highest correlation between the radiographic and US methods was found in the normal bone age group (80.9% to 86.1%) with weaker correlations for the delayed bone age group (77.1% to 86.9%) and the advanced bone age group (62.2% to 81.1%). US tended to overread delayed bone age and underread advanced bone age. US had poor positive and negative predictive value for identification of a normal or delayed bone age. The negative predictive value of US was 91% for an advanced bone age. Conclusions: On the basis of our data, US assessment should not yet be considered a valid replacement for radiographic bone age determination.