Thyroid hormone transport across the blood brain barrier in hypothyroid patients is clinically important yet poorly understood. To study this question, 200 μg of thyroxine (T4), 100 μg of 3,5,3'-triiodothyronine (T3) and 100 μg of 3,3',5'-triiodothyronine (reverse T3) were administered separately to 3 baboons, first iv and at a later date intrathecally (IT). Six animals were used. Three received the iv injections and three received the IT injections. In each of the 18 experiments, cerebrospinal fluid (CSF) and serum specimens were collected serially for 6 h after injection. Mean maximal elevations from baseline in CSF iodothyronine levels were 100 ± 10 ng/dl after iv T4, 3921 ± 293 ng/dl after iv T3 and 31 ± 17 ng/dl after iv reverse T3. When given IT in the same dosages, the mean maximal increases in serum iodothyronine concentrations were: 1670 ± 600 ng/dl for T4, 806 ± 405 ng/dl for T3, and 210 ± 43 ng/dl for reverse T3. In every animal studied, rapid bidirectional transfer of T3 from serum to CSF and CSF to serum occurred, whereas iv T4 resulted in delayed minimal increments in CSF T4 concentration. Isotopic experiments were also performed and the results analyzed using a kinetic model. When 125I-T3 was given iv, the equilibrium point in CSF was observed within 90 min with 1.7% of the administered dose/L able to be counted in CSF at any moment in time. When labeled T4 was given iv, only 0.6% of the administered dose/L was counted in CSF and the equilibrium point was not reached until 360 min. These data suggest: (a) T4, T3, and reverse T3 are all capable of bidirectional transfer across the blood brain barrier, (b) T3 may be a better agent than T4 in treating patients with myxedema coma because T3 crosses more rapidly and more completely from serum to CSF.