Hepatocyte nuclear factor 4α (HNF4α) (NR2A1), an orphan member of the nuclear receptor superfamily, binds DNA exclusively as a homodimer even though it is very similar in amino acid sequence to retinoid X receptor α (RXRα), which heterodimerizes readily with other receptors. Here, experimental analysis of residues involved in protein dimerization and studies on a reported ligand for HNF4α are combined with a structural model of the HNF4α ligand-binding domain (LBD) (residues 137 to 384). When K300 (in helix 9) and E327 (in helix 10) of HNF4α1 were converted to the analogous residues in RXRα (E390 and K417, respectively) the resulting construct did not heterodimerize with the wild-type HNF4α, although it was still able to form homodimers and bind DNA. Furthermore, the double mutant did not heterodimerize with RXR or RAR but was still able to dimerize in solution with an HNF4α construct truncated at amino acid residue 268. This suggests that the charge compatibility between helices 9 and 10 is necessary, but not sufficient, to determine dimerization partners, and that additional residues in the HNF4α LBD are also important in dimerization. The structural model of the HNF4α LBD and an amino acid sequence alignment of helices 9 and 10 in various HNF4 and other receptor genes indicates that a K(X)₂₆E motif can be used to identify HNF4 genes from other organisms and that a (E/D(X)_26–29K/R) motif can be used to predict heterodimerization of many, but not all, receptors with RXR. In vitro analysis of another HNF4α mutant construct indicates that helix 10 also plays a structural role in the conformational integrity of HNF4α. The structural model and experimental analysis indicate that fatty acyl CoA thioesters, the proposed HNF4α ligands, are not good candidates for a traditional ligand for HNF4α. Finally, these results provide insight into the mechanism of action of naturally occurring mutations in the human HNF4α gene found in patients with maturity onset diabetes of the young 1 (MODY1).