Nuclear hormone receptor (NR) signaling, currently a therapeutic target in multiple diseases, involves an ordered series of protein interactions to regulate transcription in response to changing hormone levels. Later steps in the process of ligand-dependent signaling are driven by a highly conserved interaction between the NRs and the steroid receptor coactivators (SRCs) that is effected by a conserved interaction motif (L1XXL2L3), known as an NR box. Using computational design and combinatorial chemistry, we have produced novel alpha-helical proteomimetics of the second NR box of SRC2 that exploit structural differences between human estrogen receptor alpha (hERalpha), human estrogen receptor beta (hERbeta), and human thyroid hormone receptor beta (hTRbeta). The resulting library sequentially replaced each leucine with non-natural side chains. Screening this library using a quantitative competition assay revealed compounds that selectively inhibit the interaction of SRC2-2 with each individual NR in preference to its interaction with the other NR. This approach generated highly selective compounds from one that had no specificity for a particular family member. These compounds represent the first family-member-selective competitive inhibitors of the protein interactions of transcription factors.