Structural studies on G-protein-coupled receptors have been hampered for many years by their instability in detergent solution and by the number of potential conformations that receptors can adopt. Recently, the structures of the beta(1) and beta(2) adrenergic receptors and the adenosine A(2a) receptor were determined in the antagonist-bound state, a receptor conformation that is thought to be more stable than the agonist-bound state. In contrast to these receptors, the neurotensin (NT) receptor NTS1 is much less stable in detergent solution. We have therefore used a systematic mutational approach coupled with activity assays to identify receptor mutants suitable for crystallization, both alone and in complex with the peptide agonist NT. The best receptor mutant NTS1-7m contained four point mutations. It showed increased stability compared to the wild-type receptor, in the absence of ligand, after solubilization with a variety of detergents. In addition, NTS1-7m bound to NT was more stable than unliganded NTS1-7m. Of the four thermostabilizing mutations, only one residue (A86L) is predicted to be in the lipid environment. In contrast, I260A appears to be buried within the transmembrane helix bundle, F342A may form a distant part of the putative ligand-binding site, whereas F358A is likely to be in a region that is important for receptor activation. NTS1-7m binds NT with a similar affinity for the wild-type receptor. However, agonist dissociation was slower, and NTS1-7m activated G-proteins poorly. The affinity of NTS1-7m for the antagonist SR48692 was also lower than that of the wild-type receptor. Thus, we have successfully stabilized NTS1 in an agonist-binding conformation that does not efficiently couple to G-proteins.