This paper examines several clinical concerns about the shorter half-life benzodiazepine hypnotics from an epidemiologic perspective. It draws on data from (1) 1979 and 1990 comprehensive probability-based U.S. national household surveys of the medical use of psychotherapeutic medications; (2) a 1990 four-city community-based volunteer call-in survey of the beneficial and adverse effects of hypnotics; and (3) an analogous random-digit dialing telephone survey in the general population. The issues addressed are abuse liability, rebound, depersonalization/derealization, paranoid feelings, accidents/injuries, and the unexamined consequences of the target illness in assessments of benefit-risk. In populations representative of everyday outpatient practice, we found that (1) the abuse liability of benzodiazepine hypnotics with shorter and longer elimination half-lives was generally low and comparable; (2) prevalence rates for rebound were low and not differential for flurazepam, temazepam, triazolam, and OTC sleeping pills; (3) reports of a single or an occasional experience involving depersonalization/derealization or paranoid feelings were fairly frequent in normals, in insomnia patients prior to treatment, and in persons with untreated insomnia; (4) treatment-emergent rates of occurrence for these same symptoms were low and not drug-specific; (5) past-year prevalence rates for serious accidents/injuries were much higher for chronic untreated insomnia than for normal controls and most groups treated with psychotherapeutic medications. A high proportion of past-year users of hypnotics were satisfied with their medication and would take it again.