In view of continuing debate regarding the best definitive therapy for thyrotoxicosis, we examined the long-term outcome of radioiodine (131I) or surgical treatment of 1918 thyrotoxic patients divided into three groups: those given 131I at a dose calculated from thyroid size, 131I uptake and effective half-life to administer a fixed radioactivity dose to the thyroid; those treated with a dose of 131I (110, 185 or 370 MBq) chosen empirically; and those treated by partial thyroidectomy. A minimum 10-year follow-up was achieved for 1119 patients treated with a calculated 131I dose; a single dose resulted in control of disease in 90.5%. At 5 years, 18% were hypothyroid, the prevalence rising to 42% at 20 years. Of 504 patients treated with an empirical 131I dose and followed for at least 5 years, thyrotoxicosis was controlled by a single dose in 89.7%. The rate of hypothyroidism at 5 years (38.5%) was higher than that found in the calculated dose group. A minimum 10-year follow-up was achieved for 295 surgically treated patients; thyrotoxicosis was controlled in 89.2%. The prevalence of hypothyroidism (2% at 5 years, 27.5% at 20 years) was lower than that found after 131I, whether given by calculated or empirical dose. Each of the treatments employed resulted in an acceptable rate of cure of thyrotoxicosis. If maintenance of euthyroidism is the major objective, our findings suggest that surgery represents the treatment of choice. Furthermore, calculated dose 131I administration has advantages in terms of risk of hypothyroidism over empirical dose treatment.(ABSTRACT TRUNCATED AT 250 WORDS)