Background: Decisions regarding selection and sequencing of treatment in advanced prostate cancer (PC) are often made empirically, without information on the PC molecular profile at the time of treatment. CTC provide real time information regarding contemporaneous biologically relevant cancer subtypes. Current assays often rely on EpCAM or other epithelial markers and may underestimate or exclude other CTC populations of clinical relevance. We devised a novel, simple, rapid and cheap method for isolation of CTC subsets including those without epithelial markers. Methods: Five mL of blood was processed as follows: CD45+ depletion (RosetteSep Human CD45 Depletion Cocktail, Stemcell Technologies); Ficoll centrifugation; antibody staining (mix of EpCAM-PE, PSMA-APC, N-cadherin-AF488, E-cadherin-PerCP/Cy5.5, CD45-APC-H7); Sytox Blue live/dead cell stain; followed by sorting (Becton Dickinson FACSAria Fusion cell sorter). Live cells were sorted into 4 categories denoted E, E+P, P, N (Table). N and P are not identifiable using existing conventional CTC techniques. CTC were assayed for prostate markers (AR, FOLH1 (PSMA), KLK3 (PSA), TMPRSS2, AMACR) and malignant markers (PTEN copy number) using digital droplet PCR. Results: Preliminary data are derived from 62 patient samples providing 47 sorted populations. Details are in the Table. Spiking experiments reliably identified as few as 10 CTC / 7.5 mL of blood. P cells were detected in all patients, often outnumbering sum of E and E+P; E and N were never co-expressed. P and N samples frequently expressed prostate markers. Only 1/27 showed no expression of any of these five prostate markers. PTEN deletion assays are in progress. Conclusions:“Unconventional” EpCAM^neg CTC appear to be derived from PC and can be detected using broader cell selection criteria. These CTC may provide information on treatment resistance or escape mechanisms.