A prospective longitudinal study of thyroid function was conducted in 180 consecutive admissions to an acute geriatric unit. The rest strategy included assay of TSH and TRH testing when appropriate. On admission TSH was suppressed ( < 0.1 mU/l) in 17 patients (9.4%) and elevated ( > 4.0 mU/l) in 8 (4.4%). Follow-up of these initial abnormalities showed resolution in almost all cases. Our findings suggest that acute illness may interfere with the hypothalamic-pituitary-thyroid axis, causing either temporary suppression or stimulation of TSH release. Although no new cases of thyroid disease were diagnosed, the prevalence of established dysfunction was 3.3% (2.7% hypothyroid; 0.6% hyperthyroid). To avoid misleading results, testing of thyroid dysfunction should be delayed until recovery from illness.