Earlier work has shown that plasma vasopressin levels of depressed patients were higher than those of healthy controls. The aim of the present study was to determine whether plasma vasopressin levels were correlated to parameters of the circadian rhythm. Forty-one patients with major depression and twenty-five controls participated in a case–control design under natural circumstances in a field study to investigate plasma vasopressin levels three times daily, circadian motor activity, and the 24-h periodicity of body temperature for five consecutive 24-h periods. Temperature measurements consisted of at least five, but mostly six or more measurements every 24 h. Twenty-two percent of the patients, but none of the controls lacked 24-h periodicity of body temperature. In melancholic patients increased vasopressin levels in plasma correlated with a weak 24-h periodicity of body temperature. The role of vasopressin is discussed in the light of the present findings.