Human glutamate carboxypeptidase II (GCPII) is a transmembrane metallopeptidase found mainly in the brain, small intestine, and prostate. In the brain, it cleaves N-acetyl-L-aspartyl-glutamate, liberating free glutamate. Inhibition of GCPII has been shown to be neuroprotective in models of stroke and other neurodegenerations. In prostate, it is known as prostate-specific membrane antigen, a cancer marker. Recently, human glutamate carboxypeptidase III (GCPIII), a GCPII homolog with 67% amino acid identity, was cloned. While GCPII is recognized as an important pharmaceutical target, no biochemical study of human GCPIII is available at present. Here, we report the cloning, expression, and characterization of recombinant human GCPIII. We show that GCPIII lacks dipeptidylpeptidase IV-like activity, its activity is dependent on N-glycosylation, and it is effectively inhibited by several known inhibitors of GCPII. In comparison to GCPII, GCPIII has lower N-acetyl-L-aspartyl-glutamate-hydrolyzing activity, different pH and salt concentration dependence, and distinct substrate specificity, indicating that these homologs might play different biological roles. Based on a molecular model, we provide interpretation of the distinct substrate specificity of both enzymes, and examine the amino acid residues responsible for the differences by site-directed mutagenesis. These results may help to design potent and selective inhibitors of both enzymes.