Many investigations of noise-induced hearing loss have demonstrated a poor correlation between hearing threshold and hair cell loss. One reason for this is that more subtle changes in the hair cell, such as detailed morphological changes of stereocilia, have not been evaluated. However, examining such changes increases the problem of distinguishing experimental pathological changes from artefacts. Preparation of the specimen for scanning electron microscopy (SEM) may result in too many artefacts for an adequate quantification of defects due to noise exposure. One problem with some earlier studies seems to be lack of controls and/or statistical analysis for the purpose of eliminating the influence of artefacts and spontaneous degeneration. The aim of this study was to compare unexposed and noise-exposed cochleas examined with SEM in order to determine if subtle changes due to noise could be distinguished from preparation artefacts and from spontaneous deterioration. Ten different types of hair cell changes were found in exposed and control animals. By means of using controls for statistical comparison with noise-exposed animals two cell damage categories--hair cell loss and missing stereocilia--were found to be produced by exposure to noise.