The effects of retinoic acid on cell proliferation, differentiation and patterning are thought to be mediated by the various retinoic acid receptors. Different receptor types are encoded by distinct genes (a, b, and g), whereas various isoforms within each type are encoded by splicing variants resulting from the use of alternative promoters. The only region that differs between isoforms is the N-terminal A region containing a transcriptional activating domain. It has been proposed that these alternative A regions confer distinct activities on the receptors, thus allowing each to mediate specific effects of retinoic acid, but it has been difficult to demonstrate such isoform specificity as most cells express a number of different retinoic acid receptors. In an attempt to test whether different isoforms can mediate distinct biological effects we are focusing on retinoic-acid-dependent growth inhibition of newt limb cells. We have constructed chimaeric receptors in which the retinoic acid binding domain of each of five newt retinoic acid receptors has been replaced with a thyroid hormone (T3) binding domain. These constructs were introduced individually into cells whose growth rate was then measured in the presence of T3. The chimaeric a1 receptor mediated T3-dependent inhibition of proliferation that was comparable to that given by retinoic acid, whereas the a2 isoform had no activity in this assay, nor did the d1A, d1B and d2 receptors. When the A region was deleted from the a1 chimaera it remained a potent T3-dependent transcriptional activator, but no longer mediated T3-dependent growth inhibition. In contrast, when the A region of a1 was transferred to a d chimaeric receptor, the resulting molecule was fully active in T3-dependent growth inhibition. This is the first direct evidence for isoform specificity in a biological response to retinoic acid, and demonstrates that the specificity of this response is confined to the A region.