The DIAGNOdent system is intended as an aid to caries detection and has undergone extensive investigation in experimental and clinical studies. Its applicability for epidemiological purposes has yet to be evaluated. The Objective of the present study was to evaluate the DIAGNOdent system under field conditions.The subjects comprised a sub-sample from an on-going epidemiological study. PARTICIPANTS AND CLINICAL SETTING: In 39 individuals, 97 sites were visually identified as possibly carious: 42 occlusal, 13 lingual, 19 buccal and 23 approximal sites.Three DIAGNOdent measurements were taken at each site under varying conditions: (1) naturally wet, (2) dried, and (3) polished and dried surfaces. Three sets of measurements were obtained and compared by intra-class correlation and paired sample t-tests.All correlation coefficients were > 0.9. There was a systematic difference between mean scores from naturally wet sites (without cleaning) and from either dried, or polished and dried sites. The difference was marginal and did not impact on the threshold for recording sites as carious. In two subjects, high unexpected values were recorded on sound sites prior to actual measurements, i.e. during establishment of a standard value for a sound site. The overall findings were not affected by these values, as they were consistently high in all three measurements.The DIAGNOdent system performed satisfactorily under different measurement conditions. The present study indicates that the DIAGNOdent system would be applicable in field studies, provided consistent study conditions are maintained and unexpected values are interpreted with caution.