Preclinical and clinical data have shown that D-cycloserine (DCS), a partial agonist at the N-methyl-d-aspartate receptor complex, augments the retention of fear extinction in animals and the therapeutic learning from exposure therapy in humans. However, studies with non-clinical human samples in de novo fear conditioning paradigms have demonstrated minimal to no benefit of DCS. The aim of this study was to evaluate the effects of DCS on the retention of extinction learning following de novo fear conditioning in a clinical sample. Eighty-one patients with social anxiety disorder were recruited and underwent a previously validated de novo fear conditioning and extinction paradigm over the course of three days. Of those, only 43 (53%) provided analyzable data. During conditioning on Day 1, participants viewed images of differently colored lamps, two of which were followed by with electric shock (CS+) and a third which was not (CS-). On Day 2, participants were randomly assigned to receive either 50 mg DCS or placebo, administered in a double-blind manner 1 hour prior to extinction training with a single CS+ in a distinct context. Day 3 consisted of tests of extinction recall and renewal. The primary outcome was skin conductance response to conditioned stimuli, and shock expectancy ratings were examined as a secondary outcome. Results showed greater skin conductance and expectancy ratings in response to the CS+ compared to CS- at the end of conditioning. As expected, this difference was no longer present at the end of extinction training, but returned at early recall and renewal phases on Day 3, showing evidence of return of fear. In contrast to hypotheses, DCS had no moderating influence on skin conductance response or expectancy of shock during recall or renewal phases. We did not find evidence of an effect of DCS on the retention of extinction learning in humans in this fear conditioning and extinction paradigm.