We performed a cross-over study in which triazolam was administrated for 2 weeks, followed by a 2-week treatment with zopiclone, or vice versa, in 27 patients with sleep disorders in psychiatric field. The diagnoses were schizophrenia in 10 patients, neurosis in 7, depression in 6, neurotic insomnia in 3 and organic psychosis in 1. Triazolam treatment resulted in a global improvement rate of 77.8% as compared to 63.0% for zopiclone treatment and a utility rate (“very useful”+“useful”) of 77.8% as compared to 63.0% for zopiclone treatment. Both global improvement and utility rates were thus significantly higher for triazolam than zopiclone (p < 0.01). Adverse reactions were reported by 2 patients (7.4%) during triazolam treatment and 6 patients (22.2%) during zopiclone treatment, but abnormal laboratory data were not observed. Triazolam was found to be significantly superior to zopiclone in the overall comparison made taking all above indexes into account (p < 0.01). All sleep parameters except dream showed significant improvement during both triazolam and zopiclone treatments when compared to pre-treatment baselines. The dream parameter improved significantly only during triazolam treatment as compared to the pre-treatment baseline. There were significant differences between triazolam and zopiclone regarding sleep onset and sleep soundness. These findings indicate that triazolam is superior to zopiclone as a sleep inducer in terms of clinical efficacy, adverse reactions, utility, sleep onset, sleep soundness and dream when used in the treatment of sleep disorders associated with psychiatric disorders.