BACKGROUND: Fresh patient specimens of castrate-resistant prostate cancer (CRPC) are invaluable for studying tumor heterogeneity and responses to current treatments. They can be used for primary patient-derived xenografts (PDXs) or serially transplantable PDXs, but only a small proportion of samples grow successfully. To improve the efficiency and quality of PDXs, we investigated the factors that determine the initial engraftment of patient tissues derived from TURP specimens. METHODS: Fresh tissue was collected from castrate patients who required a TURP for urinary symptoms. Tissue was grafted under the renal capsule of immune-compromised mice for up to 14 weeks. The abundance of cancer in ungrafted and grafted specimens was compared using histopathology. Mice were castrated or implanted with testosterone pellets to determine the androgen-responsiveness of CRPC PDXs from TURP tissue. RESULTS: Primary PDXs were successfully established from 7 of 10 patients that underwent grafting. Of the 112 grafts generated from these 10 patients, 21 contained cancer at harvest. Grafts were most successful when the original patient specimens contained high amounts of viable cancer, defined as samples with (i) at least 50 cancer cells, (ii) no physical damage, and (iii) detectable Ki67 expression. PDX grafts survived in castrated hosts and proliferated in response to testosterone, confirming that they were castrate resistant but androgen-responsive. CONCLUSIONS: Primary PDXs of CRPC can be established from TURP specimens with modest success. The take rate can be increased if the original tissues contain sufficient numbers of actively proliferating cancer cells. Selecting specimens with abundant viable cancer will maximize the rate of engraftment and increase the efficiency of establishing PDXs that can be serially transplanted. Prostate 9999: XX-XX, 2015. (c) 2015 Wiley Periodicals, Inc.