The aim of this study was twofold: first, to see if the prevalence of the sleep apnea syndrome (SAS) in a given population could be fairly estimated by our patient questionnaire, mainly based upon the 1979 American Sleep Association definition of SAS; and second, to investigate whether the severity of SAS could be similarly accurately measured by daytime polysomnography (DPSG), as an alternative to the more demanding all-night polysomnography (NPSG). Of 42 patients consecutively examined due to rhonchopathy, 18 had the clinical diagnosis of SAS, which was based on the three symptoms--snoring, sleep disturbances and diurnal hypersomnia--if reported to occur habitually. In 11 patients the diagnosis was established by NPSG [apnea index (AI) greater than 10]. However, in only 10 of the 18 cases NPSG indicated the diagnosis giving a positive predictive value of 56%. When comparing DPSG versus NPSG in 36 patients, the AI ranged from -23 to +65, and the mean AI value was found to be twice as high in the former (mean difference 9.0 +/- 18.4; p less than 0.01). The positive predictive value of DPSG was 63% (10/16). Both the self-report and DPSG were burdened with some 25% false-positive results, and DPSG gave far too variable AI values to be reliable in staging the disease. On the other hand, the negative predictive values were high, 96% (23/24) and 100% (20/20), respectively, indicating their usefulness for screening purposes.