RNAi (RNA interference) has been widely used to silence specific genes. However, RNAi may also cause off‐target silencing and elicit non‐specific side effects. To achieve cell‐specific gene silencing, a cell‐selective promoter has to be used to drive RNAi expression. Furthermore, different terminators of cell‐selective promoters may cause different silencing efficacies. In order to explore the best promoter and terminator combination and prove the cell‐selective gene silencing effect of PSMAe/p (prostate‐specific membrane antigen enhancer/promoter), we first constructed three plasmids by using PSMAe/p and three different terminators [poly(A), minipoly(A) and poly(U)] to explore the cell‐selective driving ability of PSMAe/p by targeting EGFP (enhanced green fluorescent protein) in LNCaP, PC‐3, EJ and HEK‐293 (human embryonic kidney) cells. Then we chose NS (nucleostemin), an important endogenous gene of prostate cancer, and constructed the NS‐targeting shRNA (small‐hairpin RNA) expression plasmid by using PSMAe/p—poly(A) combination. Cell proliferation, cell cycle and early apoptosis in vitro and xenograft tumour growth in BALB/c nude mice in vivo were detected after NS knockdown. Results showed that PSMAe/p can drive EGFP silencing in LNCaP, not in PC‐3, EJ and HEK‐293 cells and PSMAe/p—poly(A) combination achieved the best silencing efficacy. Then PSMAe/p‐shNS‐poly(A) drives NS knockdown in LNCaP cells, not in PC‐3, EJ and HEK‐293 cells. Furthermore, RNAi‐mediated NS knockdown not only reduces cell proliferation rate, reduces the percentage of S‐stage cells and increases the percentage of G1‐stage cells and increases the early apoptosis ratio in LNCaP cells in vitro, but also inhibited the LNCaP xenograft tumour growth in BALB/c nude mice in vivo by intratumoural injection. In conclusion, we have demonstrated that PSMAe/p—poly(A) combination is a promising delivery system for targeted RNAi gene therapy of prostate cancer. We showed one effective antitumour strategy by targeting NS protein, an important target in prostate cancer, with PSMAe/p‐shNS‐poly(A). These results serve as an important step for developing novel strategies to treat prostate cancer.