Snoring occurs commonly in children and is sometimes associated with obstructive sleep apnea syndrome (OSAS). Based on clinical history alone, it is difficult to distinguish primary snoring, characterized by noisy breathing during sleep without apnea or hypoventilation, from snoring indicative of OSAS. An overnight polysomnogram (PSG) is required to establish a definitive diagnosis of OSAS. Because sleep evaluations are costly and resources are limited, we evaluated whether a home audiotape recording could accurately identify children with OSAS. We studied 36 children referred by pediatricians and otolaryngologists for possible OSAS. Parents completed a questionnaire about their child's sleep and breathing and made a 15-min audiotape of the child's breath sounds during sleep. Overnight PSGs were performed on all patients. There were 29 patients who completed the study: 15 patients in the Primary Snoring group (apnea/hypopnea index < 5) and 14 patients in the OSAS group (apnea/hypopnea index > or = 5). No significant statistical differences existed between the two groups for physical characteristics or questionnaire responses. Seven observers analyzed the audiotapes for the presence of a struggle sound and respiratory pauses. The median sensitivity of the audiotape as a predictor of OSAS was 71% (range 43-86%), and the median specificity was 80% (range 67-80%). The presence of a struggle sound on the audiotape was the parameter most predictive of OSAS. There was a good level of agreement among the seven audiotape observers, as demonstrated by a mean and range kappa statistic of 0.70 (0.50-0.93) for the 21 pairs of observers. Using a clinical score to predict OSAS, the sensitivity was 46%, and the specificity was 83%. We conclude that findings on a home audiotape can be suggestive of OSAS, but are not sufficiently specific to reliably distinguish primary snoring from OSAS.