A series of aromatic and azepine ring-modified analogs of 3-hydroxy-1H-1-benzazepine-2,5-dione (HBAD) were synthesized and evaluated as antagonists at NMDA receptor glycine sites. Aromatic ring-modified HBADs were generally prepared via a Schmidt reaction with substituted 2-methoxynaphthalene-1,4-diones followed by demethylation. Electrophilic aromatic substitution of benzazepine 3-methyl ethers gave 7-substituted analogs. The preparation of multiply substituted 2-methoxynaphthalene-1,4-diones was effected via Diels-Alder methodology utilizing substituted butadienes with 2-methoxybenzoquinones followed by aromatization. Structural modifications, such as elimination of the aromatic ring, removal of the 3-hydroxyl group, and transfer of the hydroxyl group from C-3 to C-4, were also studied. An initial evaluation of NMDA antagonism was performed using a [3H]MK801 binding assay. HBADs demonstrating NMDA antagonist activity as indicated by inhibition of [3H]MK801 binding were further evaluated employing a [3H]-5,7-dichlorokynurenic acid (DCKA) glycine site binding assay. Selected HBADs were characterized for functional antagonism of NMDA and AMPA receptors using electrophysiological assays in Xenopus oocytes and cultured rat cortical neurons. Antagonist potency of HBADs showed good correlation between the different assay systems. HBADs substituted at the 8-position possessed the highest potency with the 8-methyl (5), 8-chloro (6), and 8-bromo (7) analogs being the most active. For HBAD 6, the IC50 in [3H]-DCKA binding assays was 0.013 microM and the Kb values for antagonism of NMDA receptors in oocytes (NR1a/2C) and cortical neurons were 0.026 and 0.048 microM, respectively. HBADs also antagonized AMPA-preferring non-NMDA receptors expressed in oocytes but at a lower potency than corresponding inhibition of NMDA receptors. HBADs demonstrating a high potency for NMDA glycine sites showed the highest steady-state selectivity index relative to AMPA receptors. Substitution at the 6-, 7-, and 9-positions generally reduced or eliminated glycine site affinity. Moving the hydroxyl group from C-3 to C-4 reduced receptor affinity, and potency was eliminated by the removal of the aromatic ring or the hydroxyl group. These data indicate that the HBAD series has specific structural requirements for high receptor affinity. With the exception of substitution at C-8, modified HBADs generally have a lower affinity at NMDA receptor glycine sites than the parent compound 3. Mouse maximum electroshock-induced seizure studies show that the three HBADs selected for testing have in vivo potency with the 6,8-dimethyl analog (52) being the most potent (ED50 = 3.9 mg/kg, iv).