The neural basis of generalized anxiety disorder (GAD) is poorly characterized. The effect of chronic administration (12 weeks) of paroxetine, a selective serotonin reuptake inhibitor, on N-acetylaspartate (NAA), a marker of neuronal viability, was evaluated in adults with GAD using proton magnetic resonance spectroscopic imaging ((1)H MRSI) at 1.5 T. We hypothesized that, pretreatment abnormalities in hippocampal NAA/creatine (NAA/Cr) would normalize with symptomatic improvement. Nine GAD patients (mean age = 41.7 year; 4 females) received 12 weeks of open-label paroxetine treatment, flexibly dosed up to 60 mg/day. Clinical outcome was assessed with the Hamilton Anxiety Rating Scale (HAM-A). Multislice ( 1)H MRSI scans were performed at unmedicated baseline and following 6 and 12 weeks of treatment. Ten untreated healthy volunteers (HVs) (mean age = 37.1 year; 4 females) received scans at the same intervals. All patients achieved remission (HAM-A <or= 7) by week 12. Compared to HVs, GAD patients showed persistently lower levels of bilateral hippocampal NAA/Cr (17.7% mean decrease; Cohen's d = 1.29) that were maintained across all three time points, despite marked symptom improvement. This pilot study failed to support an association between a hippocampal neuronal marker and anxiolytic response to paroxetine, and suggests further investigation of potential trait-like hippocampal abnormalities in GAD.