A randomized, controlled, crossover clinical study compared 14-night treatment with 15 mg temazepam, 50 mg diphenhydramine, and placebo in elderly individuals with insomnia (mean age, 73.9 years; range, 70-89 years). Primary outcome measures were subjective assessments of sleep recorded on sleep diaries. Secondary measures were the morning-after psychomotor impairment, using the digit symbol substitution task and the manual tracking task, and the morning-after memory impairment, using a free-recall procedure. Results showed sleep improvements with 15 mg temazepam compared with placebo-sleep quality (mean score, 3.3 +/- 0.9 vs 2.9 +/- 0.8; P = 0.03), total sleep time (6.9 +/- 1.0 hours vs 6.3 +/-1.3 hours; P = 0.02), number of awakenings (1.5 +/- 1.3 vs 2.0 +/- 1.2; P < 0.001), and sleep-onset latency (25 +/- 22 minutes vs 37 +/- 25 minutes; P = 0.03). Improvements were seen with diphenhydramine treatment compared with placebo on the number of awakenings only (mean, 1.7 +/- 1.1 vs 2.0 +/- 1.2; P < 0.05). Numbers of adverse events reported were similar after all treatments, although there was 1 fall during temazepam treatment. Findings indicate that temazepam is more effective than diphenhydramine when compared with placebo at the doses tested, although this advantage is mitigated by the risk of falls associated with temazepam use. The choice of agent to use in the elderly must consider these relative benefits and risks.