In France, young men aged 18‐24 years undergo systematic medical assessments in the selection centres of the army. We did a cross-sectional study of risk factors for deafness in 1208 young men from two such selection centres in Vincennes (n=525) and Lyon (n=683). We assessed hearing status by tonal automated audiometry 1 at 0·5 kHz, 1·0 kHz, 2·0 kHz, 4·0 kHz, and 8·0 kHz. Hearing loss was expressed in dB relative to hearing level. The men were interviewed about hearing status and exposure to potential harmful sources of noise. Additionally, in Lyon, we recorded antecedents of childhood ear diseases, including history of repeated episodes of otitis media. We checked that the men were not exposed to noise before audiometry. Hearing thresholds in exposed and non-exposed groups were compared by ANOVA. Of the total population, in 449 (38%) of 1194, the hearing threshold at 1·0 kHz was slightly increased (p=0·04) in men who went to rock concerts and discos twice a month or more. Hearing thresholds at 2 kHz and 4 kHz were significantly increased in (p=0·02) 214 (18%) of 1208 men with noisy occupations (for 6 months or longer). The major factor influencing hearing was use of personal stereos. Hearing thresholds were increased at all frequencies (p=0·001) in 205 (17%) of 1208 men who used personal stereos for at least 1 h per day. The difference of mean hearing loss between personal stereo users and non-users was 3·5 dB relative to hearing level. However, further analysis of the data from the Lyon centre, showed that this harmful effect was strongly dependent on the presence of absence of repeated episodes of otitis media in infancy or childhood (figure). In this centre, the 114 (17%) of 683 men who had had repeated otitis media episodes were affected. Personal stereo use had no apparent effect on the hearing status of men without previous otitis media. Use of personal stereos was associated with a mean extra hearing loss of 11 (SE 2·99) dB relative hearing loss. This difference of susceptibility to personal stereo use was significant (p=0·001). We found also increases in risks of acute acoustic trauma and frequent tinnitus (odds ratio 3·32 [95% CI 1·47‐7·50 and 5·28 2·88‐9·65], respectively). Previous studies have shown harmful effects of personal stereo use on hearing performances in young adults. 2‐5 These effects were, however, apparently of limited clinical importance (mean differences of hearing loss about 3 dB). We show that personal stereo use has clinically important effects in men with antecedents of repeated episodes of otitis media in childhood. Our results show the importance of curing otitis in childhood and of a prevention policy. Young people with a history of repeated episodes of otitis media should be warned that personal stereo use is a risk factor for deafness. This study was supported by the French Government DSP/DGA “Committee on Noise Exposure”. We thank the selection centres’ staff for their participation, L Pellieux, IMASSA, W Durand, and P Castella for their crucial technical help in informatics and acoustics.