The observations comprise a systematic study of the use of 26 laboratory and 3 radiological investigations in routine management of 1007 admissions to a District General Hospital adult psychiatry unit over 16 months. For laboratory investigations, 60% of admissions had an average of 14 results each. Results were frequently of no clinical value. Most of the 10.2% abnormal could be expected from clinical findings. Many unexpected abnormal were false positives. The percentage abnormal and the percentage abnormal accompanied by positive clinical findings varied widely between test types. From this survery, the only tests of value for routine case-finding among psychiatric admissions aged 16-65 years are thyroid function in females and urinalysis in females. Tests with special indications are white cell count in unremitting episodes of psychiatric illness and syphilis serology when previous disinhibited behaviour has exposed to risk of infection. For radiology, examination of the chest was only worthwhile in the presence of the usual medical indications including alcoholism or with unremitting periods of mental illness, and CT head scan when there was evidence of organic disease above the foramen magnum. There were no indications for skull x-ray. Radiology was compared with the findings of a survey 2 years earlier. Chest and skull x-rays had decreased and CT head scans increased. The percentage abnormal had increased for each. Mental illness attributable to unsuspected physical disease detected by investigations was rare, confined to 3 cases of thyroid dysfunction and 12 possible cases of chest infection with abnormal white cell count or x-ray report. Physical disease attributable to mental illness was common, but rarely appreciable and usually apparent clinically. Use of investigations could be improved by adoption of these recommendations and by improving clinicians' understanding of the derivation of reference ranges, combined with use of a selection of alternative ranges or action limits.