There is little published literature on the correlation between subjective and objective efficacy of hypnotics. We wanted to determine whether there was a correlation between the patient's subjective evaluation of the efficacy of the hypnotic with the polysomnographic (PSG) findings. We studied 16 patients with chronic insomnia (sleep latency, greater than or equal to 30 minutes; total sleep time, greater than 240 but less than 420 minutes) for 11 nights who took placebos on nights 1 and 2, zolpidem (imidazopyridine) on nights 3-9 and placebo on nights 10 and 11. Patients completed a questionnaire each morning following PSG, which evaluated subjective sleep quality, sleep latency and total sleep time. These data were compared to PSG findings to answer specific questions about sleep latency reduction, efficacy of the hypnotic after a week's use, sleep quality after discontinuing the drug, and any correlation between subjective and objective measures. PSG findings indicated a shortened sleep latency, increased total sleep time, decreased total wake time and increased sleep efficiency when patients ingested zolpidem 30 minutes before bedtime. We found that after 7 nights (nights 3-9) the drug was still effective in reducing sleep latency and increasing total sleep time. Upon withdrawal (nights 10 and 11) sleep returned to baseline (nights 1 and 2). Subjectively, the patients confirmed those findings on the questionnaire, as well as a subjective reduction in the number of awakenings and, interestingly, a subjective increase in the time spent awake after sleep. Many of the objective variables we examined correlated highly with the subjective variables. While on zolpidem, subjects believed and were objectively shown to have a decreased sleep latency, increased total sleep time and decreased time awake before persistent sleep, although they tended to overestimate sleep latency and time spent awake before persistent sleep and underestimated total sleep time. Although the correlation between objective and subjective measures was high for the group, in individual patients there was an impressive difference between the two, and the highest coefficient of variation between a subjective and objective measures was 0.453. No correlations were found with subjective measures of refreshing quality of sleep, decrease in number of awakenings, how sleepy patients felt in the morning or their ability to concentrate in the morning. Thus, we believe the PSG remains the keystone in the evaluation of hypnotic efficacy.