OBJECTIVE: To determine how the referral pattern from school screening of a group of dentists changed following a training programme. DESIGN: Eighty clinical photographs of teeth were shown to a group of dentists both before and after a training programme. The photographs were of teeth in various stages of the carious process from children aged between three and 15 years. The dentists were shown each of the 80 photographs and were asked to categorise each slide according to the decision they were most likely to make during school screening. CLINICAL SETTING: An NHS Community Trust in Merseyside, UK. PARTICIPANTS: A group of 12 community dental officers. OUTCOME MEASURES: Evaluation was carried out in relation to consensus refer/not refer decisions. Consensus was defined as the majority decision provided that no more than two dentists disagreed with the majority. RESULTS: As a consequence of training there was a change from no consensus to consensus in 21.5 per cent of the slides, whereas only 2.5 per cent changed from consensus to no consensus. However, after training there was still no consensus on 30 per cent of the slides. CONCLUSIONS: This study demonstrates a means by which referral decisions from school screening can be audited.