Small atrophic prostate cancers on needle biopsy are rare and difficult to distinguish from benign atrophy on needle biopsy. We report on a study of 23 needle biopsy specimens with small foci of atrophic prostate cancer from the consult service of one of the authors. In 19 cancer cases the atrophic component was pure; in 4 cases it was dominant with a minor (<5%) nonatrophic cancer component. These atrophic cancers and 16 cases of florid benign atrophy on needle biopsy were examined by immunohistochemistry for alpha-methylacyl-CoA-racemase (AMACR). All cases of cancer and atrophy were verified immunohistochemically with antibodies to basal cells (34betaE12 and p63). AMACR staining were scored as 1+ (5% to 25% of glands expressing AMACR), 2+ (26% to 50% of glands expressing AMACR), or 3+ (>50% of glands expressing AMACR). Positive staining was defined as staining above that of surrounding benign glands. AMACR was expressed in 69.6% of atrophic prostate cancers (3+, 11 cases; 2+, 3 cases; 1+, 2 cases); 30.4% (7 cases) of atrophic prostate cancer exhibited no AMACR expression. In the 4 cases with a few glands of ordinary (nonatrophic) prostate cancer, the nonatrophic cancer demonstrated more intense and a greater extent of AMACR staining. Fourteen cases (87.5%) of benign atrophy showed no AMACR expression. In 2 cases (12.5%) of benign atrophy, background immunostaining made it difficult to assess AMACR expression. We conclude that AMACR immunostaining alone is not sufficiently discriminatory in the differential diagnosis of atrophic prostate cancer versus benign atrophy. Atrophic prostate cancers are not as frequently or as strongly positive as ordinary prostate cancer. Using a panel of immunostains including AMACR, 34betaE12 and p63 (positive AMACR immunostaining along with negative basal cell markers) is recommended in the differentiation of atrophic prostate cancer and benign atrophy.