Digital noise maps produced according to the European Environmental Noise Directive (END) could provide valuable exposure information in noise and health research. However, their usefulness in epidemiological studies has not been evaluated. The objective of this study was to apply and evaluate Swedish END maps for assessments of residential traffic noise exposure. END maps from three Swedish cities were used to assess residential traffic noise exposure for a population sample of 2496 men and women included in a national Environmental Health Survey. For each subject, we assessed noise levels manually and automatically at three geographical points, using survey data to locate dwellings within buildings. Cohen's kappa coefficient (κ) was used to assess agreement between the noise estimates. To evaluate the maps, we compared the observed and predicted proportions of annoyed residents as a function of noise exposure using survey data and already established exposure-response relationships. The root mean square deviation (r.m.s.) was used to assess the precision of observed estimates. The agreement between the noise estimates ranged from κ=0.4 to 0.8. Generally, there was a high correspondence between observed and predicted exposure-response relationships for noise annoyance, regardless of method and if data on dwelling location within building were used. The best precision was, however, found when we manually corrected the noise level according to the location of the dwelling within buildings (r.m.s.=0.029). Noise maps based on the END appear useful for assessing residential traffic noise exposure, particularly if combined with survey data on dwelling location.