As loss of DNA-repair proteins is common in urothelial carcinoma (UC), a rationale can be made to evaluate the activity of poly (ADP-ribose) polymerase (PARP) inhibitors to exploit synthetic lethality. We aimed to preclinically evaluate a PARP inhibitor, CEP-9722, and its active metabolite, CEP-8983, in UC. The activity of CEP-8983 was evaluated using a 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay against human UC cell lines. Flow cytometry, COMET assay, and western blot were performed to assess apoptosis, DNA damage, and DNA-repair proteins, respectively. RT4 xenografts received placebo or CEP-9722 (100 or 200 mg/kg/day) orally. Xenografts were subjected to immunohistochemistry for apoptosis [cleaved caspase (cc)-3] and angiogenesis (CD31). CEP-8983 (1 μmol/l) reduced the viability of RT4 and T24 cells by 20%, but did not reduce the viability of 5637 and TCC-SUP cells. Apoptosis and necrosis occurred in 9.7 and 9.1% of RT4 and 5637 cells, respectively. RT4 cells showed greater DNA damage compared with 5637 cells. Increased DNA damage occurred with combination versus CEP-8983 or cisplatin alone in RT4 and 5637 cells. T24 and RT4 showed the least RAD51 foci 8 h following radiation, whereas TCC-SUP and 5637 robustly induced RAD51 foci. CEP-9722 showed dose-dependent antitumor activity in RT4 xenografts; 200 mg/kg daily was better than control (P=0.04) and 100 mg/kg was not (P=0.26). Immunohistochemistry of xenografts showed a significant increase in cc-3 and decrease in CD31 with both doses (P<0.05). Biomarker-driven evaluation of PARP inhibitors in UC is justified as the activity of CEP-9722 correlated inversely with homologous recombination repair response to DNA damage.