The preparation and binding affinity of a series of tetrahydroisoquinoline carboxylic acids at the N-methyl-D-aspartate (NMDA) subtype of the glutamate receptor is described, together with a molecular modeling analysis of NMDA agonists and antagonists. Using published NMDA ligands, the active analogue mapping approach was employed in the generation of an agonist pharmacophore model. Although known competitive antagonists such as CPP (1) could be superimposed onto the agonist model, to overcome the assumption that they bind to the same receptor site, an independent modeling approach was used to derive a separate pharmacophore model. Development of a competitive antagonist model involved a stepwise approach that included the definition of a preferred geometry for PO3H2-receptor interactions, multiple conformational searches, and the determination of volume and electronic tolerances. This model, which is described in detail, is consistent with observed affinities of potent NMDA antagonists and has provided an explanation for the observed periodicity in affinities for the known antagonists AP5, AP6, and AP7. The features of the agonist and antagonist models are compared, and hypotheses advanced about the nature of the receptor interactions for these two classes of compounds. The pharmacophore models reported herein are consistent with a single recognition site at the NMDA receptor that can accommodate both agonist and antagonist ligands. To assist in first defining and later exploring the predictive power of the competitive antagonist model, a series of conformationally constrained NMDA antagonist (phosphonoalkyl)tetrahydroisoquinoline-1- and 3-carboxylates was prepared. From this work, 1,2,3,4-tetrahydro-5-(2-phosphonoethyl)-3- isoquinolinecarboxylic acid (89) was identified as the most active lead structure, with an IC50 of 270 nM in [3H]CPP binding. The synthesis and structure-activity relationships of these novel antagonists are described.