Obstructive sleep apnea (OSA) may induce psychiatric problems, but clinical risk factors do not reliably predict laboratory-verified OSA. Therefore, OSA diagnosis requires laboratory sleep monitoring. To find additional clinical features that would sharpen indications for sleep monitoring, we applied univariate analyses to clinical data for 137 OSA patients seen in a psychiatry sleep clinic. A symptomatology questionnaire was obtained from 101 of these patients: 71 had morning and 86 had afternoon vigilance tests, and all had upper-airway evaluation and polysomnography. Cigarette consumption but no other clinical features differed among OSA severity groups and total sleep period groups; upper-airway findings differed among vigilance groups. Multidiscriminant clinical predictor terms categorized several patients with severe OSA into less severe OSA categories. Clinical features did not accurately predict OSA. OSA will continue to be identified primarily by sleep laboratory testing. A two-phase laboratory routine, reserving full laboratory testing for patients with negative results on initial, less expensive screening tests, might conserve resources.