Thyrotrophin (TSH) secretion was studied in 63 patients with Cushing's syndrome (53 patients with pituitary dependent Cushing's disease, eight with adrenocortical tumours, and two with the ectopic ACTH syndrome). Prior to treatment, TSH response to 200 micrograms of TRH intravenously was significantly decreased compared to controls; TSH response was 'flat' (increment less than 2 mU/l) in 34 patients (54%). Patients with a flat response to TRH had significantly higher morning and midnight cortisol levels than patients with a TSH response of 2 mU/l and more; this was not due to differences in serum thyroid hormone levels. Basal TSH, TSH increment after TRH, and stimulated TSH value, but not serum triiodothyronine, were correlated with cortisol measurements (0800 h serum cortisol, midnight cortisol, and urinary free corticoid excretion). After exclusion of 40 patients with additional disease (severe systemic disease, diabetes mellitus, or goitre), cortisol-TSH correlations were even more pronounced (r = -0.73 for midnight cortisol and stimulated TSH levels), while in the patients with additional complications, these correlations were slight or absent. Successful treatment in 20 patients was associated with a rise in thyroid hormone levels and the TSH response to TRH. These results indicate that (1) the corticoid excess but not serum T3 is the principal factor regulating TSH secretion in Cushing's syndrome, (2) a totally flat response to TRH is rare, and (3) TSH suppression and lower than normal serum thyroid hormone levels are reversible after treatment. Since factors like severe systemic disease, diabetes mellitus and goitre also affect TSH secretion, they tend to obscure the statistically significant correlations between cortisol excess and TSH secretion.