OBJECTIVE: To assess a systematic intervention in cases of delirium in elderly inpatients. DESIGN: Randomized, controlled trial. SETTING: University-affiliated, primary acute care hospital. PATIENTS: Patients aged 75 years or over admitted to the medical department. They were screened within 24 hours after admission, and 88 patients with delirium (according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, third revised edition) were detected and enrolled in the trial. The patients were randomly allocated to the treatment group (42) or the control group (46); all were followed up until the end of the study. INTERVENTION: Patients were assessed on enrolment and 1, 2, 4 and 8 weeks later. Those in the treatment group received a consultation by a geriatric internist or psychiatrist and follow-up by a liaison nurse. Those in the control group received regular medical care. OUTCOME MEASURES: Short Portable Mental Status Questionnaire (SPMSQ), Crichton Geriatric Behavioural Rating Scale (CGBRS), use of restraints, length of hospital stay, discharge to a setting providing more care than was needed before admission and mortality rate. RESULTS: Two weeks after admission, patients in the treatment group showed an improvement in their mean SPMSQ scores, from 8.2 (standard deviation [SD] 1.9) to 7.9 (SD 2.5), whereas the control group showed a deterioration, from 8.4 (SD 1.7) to 9.1 (SD 1.1); this difference had disappeared by the end of the 8-week period (p < 0.05). Mean CGBRS scores were higher in the treatment group (32.0 [SD 8.6]) than the control group (28.5 [SD 9.4]) on enrolment and had improved more markedly by the end of the 8-week period (to 23.9 [SD 7.8] v. 25.0 [SD 7.0], p = 0.06). There was no statistically significant difference between the groups in use of restraints, length of hospital stay, discharge to a setting providing more care than was needed before admission or mortality rate. CONCLUSION: The beneficial effects of systematic detection and intervention in cases of delirium in elderly inpatients were small.