SummaryBackgroundA screening ultrasonographic examination of yearling Thoroughbred horses has been described which can be performed on farm, as an adjunct to pre‐sale yearling radiography.ObjectivesTo compare ultrasonographic and radiographic findings in young Thoroughbred horses in the detection of osteochondrosis.Study designDescriptive and comparative study using ultrasonography and radiography.MethodsForty‐four horses with a mean age of 10 months underwent an ultrasonographic screening protocol of joint groups including the bilateral stifle, tarsal, carpal and metacarpophalangeal/metatarsophalangeal joints. Concurrent standardised radiographic projections of all the joints were also undertaken. A cohort of 26 of these horses had a second set of radiographs taken for yearling sales approximately 6 months later. Two reviewers reviewed all images for presence or absence of osteochondrosis lesions. Comparison of the ultrasonographic and radiographic findings was undertaken using Gwet's AC1 and the Landis and Kock scale.ResultsThe principal observer identified 48 ultrasonographic lesions and 15 radiographic lesions on the initial examination. Overall, there was agreement in screening for osteochondrosis using ultrasonographic and radiographic lesions, however, with varying levels of agreement for different joint groups. Agreement for individual joint groups was almost perfect for the carpal (0.98) and metatarsophalangeal joints (0.81), substantial for the stifle joints (0.67), moderate for the metacarpophalangeal joint (0.59) with only fair agreement of the tarsal joints (0.30). Six osteochondral lesions were recorded on the second set of radiographs. There was overall substantial agreement between the ultrasonographic findings and the second set of radiographs. A second observer reviewed the ultrasonographic and radiographic images. There was moderate interobserver agreement comparing ultrasonographic images, and almost perfect interobserver agreement comparing radiographic images.Main limitationsSmall number of horses who presented for yearling sale radiography for comparison. One ultrasonographer acquired all images, and two observers reviewed the images.ConclusionUltrasonographic screening of yearling horses with a relatively low number of lesions showed agreement with concurrent and yearling sale radiographs.