Atypical intraductal proliferation (AIP) of the prostate is characterized by morphologic features exceeding that of high-grade prostatic intraepithelial neoplasia but not meeting strict diagnostic criteria for intraductal carcinoma. We examined the clinical significance of AIP in biopsy specimens. Patients with AIP diagnosed on biopsy were identified from surgical pathology archives. Initial biopsies, any repeat biopsies, and any radical prostatectomy (RP) slides were rereviewed. We also identified a control group of 50 consecutive patients with available prostate biopsies showing invasive prostatic adenocarcinoma but no AIP and having paired RP for comparison. Medical records were searched for nonsurgical treatment and clinical outcome status. Patients with initial biopsies showing invasive adenocarcinoma with either grade group (GG) ≥3 and/or unfavorable histology (as recently defined) were excluded from both the study and control groups. Correlation with subsequent adverse pathology at rebiopsy or RP, as defined by separate criteria: unfavorable histology, large cribriform/intraductal carcinoma, GG ≥3, pN1, and/or pM1, was assessed for both groups. Phosphate and tensin (PTEN) homolog and ETS-related gene (ERG) immunohistochemistry were performed on biopsies with available paired RP, using standard protocols. One hundred forty-two patients with AIP met inclusion criteria. At initial biopsy, 16 patients (11.3%) had AIP without concomitant invasive carcinoma, whereas 126 (88.7%) also had invasive adenocarcinoma. Of the 126 invasive tumors with AIP meeting study criteria, 19 (15.1%) were GG 1 and 107 (84.9%) GG 2. One hundred thirty-nine of 142 patients with AIP (97.9%) had available clinical follow-up (mean: 36.9 mo). Fifty-two (36.3%) patients with AIP underwent RP, 36 (25.4%) had brachytherapy, 28 (19.7%) had radiotherapy, 17 (12%) remained on active surveillance, 2 (1.4%) had cryoablation, 2 (1.4%) received androgen deprivation therapy, and 1 (0.7%) had high-intensity focused ultrasound. Forty-seven of 52 patients undergoing prostatectomy (90.3%) had glass slides available for review: 30 (63.8%) were GG2, 13 (27.7%) GG3, 1 (2.1%) GG4, and 3 (6.4%) GG5. Seventeen (36.2%) patients were staged as pT2, 25 (53.2%) pT3a, and 5 (10.6%) pT3b. Forty-two of 47 (89.4%) patients had associated unfavorable histology on prostatectomy, including 41 (87.2%) with large cribriform/intraductal carcinoma, 17 (36.2%) GG≥3, and 5 (10.6%) with metastatic disease. In the 36 AIP lesions examined for PTEN and ERG immunoreactivity, 14 (38.9%) had concomitant PTEN loss and ERG over-expression, 6 (16.7%) showed PTEN loss only, and 6 (16.7%) had ERG overexpression only. AIP morphology was more predictive of risk for unfavorable histology at RP than PTEN/ERG immunophenotype. Seventeen patients not undergoing RP had rebiopsy, of which 5 (29.4%) had at least one adverse feature identified on repeat biopsy. Nineteen of 50 patients (38%) in the non-AIP control group had adverse pathology at RP (by any definition), compared with 89.4% in the AIP study group (P < 0.0001). In conclusion, AIP in prostate needle core biopsy is strongly associated with unsampled adverse pathology, defined by unfavorable histology and other traditional definitions of aggressive disease. For optimal patient risk stratification and active surveillance management, AIP should gain better recognition as a standard reporting element given its association with an increased likelihood of unsampled high-risk disease.