In this study we have demonstrated that specific binding sites for 3,5-di-iodo-l-thyronine (3,5-T2) can be detected in rat liver mitochondria. After incubation with the homogenate, liver mitochondria bound only a small portion of [3,5-125I]T2. The addition of a 100-fold excess of unlabelled 3,5-T2 caused the displacement of on average 50-60% of the [3,5-125I]T2 bound. Specific binding of 3,5-T2 to rat liver mitochondria occurred rapidly; a maximum was achieved after 5 min. Maximal binding was obtained at 37 °C, while at 0 °C and 20 °C the values were only slightly lower. Binding was maximal at pH 70; mean (±s.e.m.) values for the apparent association constant and the binding capacity (calculated at pH 7·0, 0 °C and after 30min of incubation) were 0·5±0·04×108 m -1 and 0·4±0·04 pmol/mg mitochondrial protein respectively. The specificity of binding, examined in competition studies, followed the order: 3,5-T2>3,3′-di-iodo-l-thyronine>3′,3,5-tri-iodo-l-thyronine>thyroxine. Other iodothyronines (3′,5′-di-iodo-l-thyronine, 3,5-di-iodo-d-thyronine, 3,3′, 5′-tri-iodo-l-thyronine, 3-iodo-l-thyronine and 3,5-di-iodothyroacetic acid) showed little or no competition. This suggests that the specific 3,5-T2 binding sites could be of biological relevance with respect to the understanding of the mechanism of physiological action of thyroid hormones at the cellular level.