Objective To evaluate the therapeutic effect of chemotherapeutic drugs on pancreatic carcinoma based on patient-derived xenograft (PDX) models, and to screen an individualized treatment strategy. Methods Fresh human pancreatic carcinoma tissues were subcutaneously transplanted into nude mice to establish PDX models which could be stably passaged. The traceability of PDX models was determined by STR analysis. The PDX models were treated with three different clinical chemotherapeutic drugs oxaliplatin, gemcitabine and irinotecan, respectively, and the tumor volumes were measured at different times. The therapeutic effect of those drugs was assessed by TGD mathematical model and plasma CA19 -9 test. Results The traceability of patient-derived xenograft samples was up to 99. 99%. Compared with the control group, the treatment with irinotecan and gemcitabine inhibited tumor growth significantly ( P = 0.001), and gemcitabine had even better result . The minimum toxic effect in the mice was induced by irinotecan treatment, followed by gemcitabine treatment. Conclusions Pancreatic carcinoma PDX models are successfully established and can be stably passaged. Gemcitabine shows the most inhibitory effect on tumor growth based on TGD mathematical model assessment, and deserves to be recommended as the preferred drug for individual treatment of pancreatic carcinoma.