We evaluated the diagnostic usefulness of the 34betaE12-p63 cocktail, compared with 34betaE12 and p63 used alone, in 34 prostate needle biopsy (NBXs) and 3 transurethral resection specimens containing atypical glandular proliferations and in 18 NBXs containing unequivocal prostate carcinoma (PCa). Staining intensity; percentage of basal cells staining in benign, atypical, and malignant glands; number of benign glands lacking basal cell staining; and staining variance were analyzed. All NBXs with unequivocal PCa were negative with all 3 markers. Diagnoses were as follows for the atypical cases after staining for the 3 markers: PCa, 9; postatrophic hyperplasia, 12; high-grade prostatic intraepithelial neoplasia (HGPIN), 5; atypical adenomatous hyperplasia, 6; benign atypical proliferations, 4; and HGPIN with adjacent small atypical acinar proliferation suggestive of PCa, 1. The cocktail demonstrated consistently strong staining intensity and improved basal cell staining in morphologically benign and benign atypical glands compared with p63 and 34betaE12 alone; it had the smallest staining variance compared with 34betaE12 (F < 0.0001) and p63 (F = 0.31), although its advantage for resolving individual atypical cases was limited compared with 34betaE12 and p63 alone. Of 37 atypical cases, 1 (3%) additionally was resolved as benign using the cocktail and p63. Because the diagnosis of PCa is supported by lack of basal cell staining, the immunohistochemical analysis with highest possible sensitivity and lowest possible variability is critical to ensure that a negative reaction is true. The cocktail provides a simple, cost-effective improvement in basal cell immunohistochemical analysis of difficult prostate lesions.