Diurnal variations are reported in pain perception, potency of placebo and in the pharmacokinetics of morphine. It is not established if such diurnal variation should be reflected in the routine prescription of analgesic drugs to cancer patients. One approach for investigating this question is to study the circadian rhythm of patients' requests for analgesics. We included 40 patients with malignant disease and intolerable pain despite treatment with weak opioids. Weak opioids were stopped and the patients' pain treatment was restricted for the next 2 days to strong oral opioids (ketobemidone) given on an as-needed basis. The number and times of all ketobemidone administrations were recorded. No significant diurnal variation in use of ketobemidone was observed. A descriptive curve that interpolates between successive observations displayed only slight circadian fluctuations of opioid consumption. Thus, the data reported in this paper does not support the necessity of considering chronopharmacological factors in the routine prescription of oral opioids.