The theory that poorly differentiated prostate carcinoma develops a neuroendocrine (NE) phenotype is controversial. Supportive data is variable with NE expression being observed in anywhere from 5% to 83% of prostate cancers. These percentages are derived from standard immunohistochemistry studies, which make no attempt to quantify the results. High-density tissue microarrays (TMAs), represent a novel method for evaluating up to 1000 tissue samples with a 0.6 mm diameter on a single glass slide. This high throughput technology for screening antibodies, however, requires validation to determine if TMAs are useful in evaluating heterogeneously expressed proteins such as the NE markers chromogranin A (CGA) and synaptophysin (SYN). This study compares results from standard slides to TMAs in 50 primary and metastatic prostate tumors taken from 12 rapid autopsies from men with hormone refractory prostate cancer. One hundred standard and 2 TMA slides were immunostained for CGA and SYN. Using standard slides, focal NE expression was seen in 1/12 primary prostate tumors. Overall, 13/100 (13%) standard slides showed focal NE expression for both primary and metastatic prostate tumors; NE expression was observed in 4/12 autopsy cases (33%) when all tumor sites per case were considered. 458 tissue elements (tumor and normal) were arrayed into one paraffin block. Seventy-three percent (332/458) of the elements placed into the TMA were confirmed histologically to represent tumor. Seventy-five percent (250/332) and 66% (218/332) could be evaluated for CGA and SYN expression, respectively. Six of the metastatic tumors expressed CGA and SYN or 2.4% (6/250; 95% CI = 0.9% to 5.2%) and 2.3% (6/218; 95% CI = 0.8% to 5.3%), respectively. In conclusion, only focal NE expression was observed by both methods (eg, standard and TMA slides). The focal expression in these advanced prostate tumors was unexpected given data from prostate tumor cell lines and animal models suggesting that progression to the NE phenotype parallels tumor progression. This study also supports the use of high density TMAs to screen for protein expression, even when expression is focal.