OBJECTIVE To evaluate the degree of agreement in first-trimester nasal bone assessment in a group of sonographers before and after training, using a semi-quantitative scoring system. METHODS Four sonographers who routinely perform first-trimester screening were first shown 46 images from both normal and trisomy 21 pregnancies. For each image, they were asked to score from 0 (disagree) to 3 (agree) on five different criteria that were deemed important in nasal-bone assessment, including image size, plane and visibility of nasal bone. A training program was then conducted, and a repeat exercise was carried out using the same 46 images. Finally, in a third exercise, images from 42 patients were presented, some having more than one image. The sonographers were required to give one overall nasal-bone score for each patient. In each exercise interobserver agreement was evaluated by intraclass correlation coefficient (ICC). RESULTS Before training, the sonographers agreed reasonably well on the five proposed criteria (ICC, 0.752), with some disagreement on their perceived image quality. The training program further improved the agreement (ICC, 0.790), particularly on whether the nasal bone was the biggest and brightest echogenic component. Agreement was excellent when they were asked to give one overall score on the nasal bone based on multiple images from one patient (ICC, 0.929). CONCLUSION The proposed scoring system can be used to improve consistency and reliability in first-trimester nasal-bone assessment.