Arsenic trioxide (ATO) has been shown to inhibit pancreatic cancer (PC) cell growth in vitro and to promote the inhibitory effects of gemcitabine (Gem) on PC in vivo. However, the high toxicity of ATO associated with the required high doses and indiscriminate targeting has limited its clinical application. This study aimed to determine whether coupling arsenic to a tumor homing peptide would increase the inhibitory potency against PC cells. The effects of this peptide-linked arsenic compound (PhAs-LHP), the analogous non-targeting arsenic compound (phenylarsine oxide, PAO), and marketed ATO on PC growth were tested in vitro and in a mouse model. The data demonstrated that PhAs-LHP inhibited PC cell growth in vitro more potently, with IC50 values 10 times lower than ATO. Like ATO, PhAs-LHP induced cell death and cell cycle arrest. This cytotoxic effect of PhAs-LHP was mediated via a macropinocytosis-linked uptake pathway as amiloride (a macropinocytosis inhibitor) reduced the inhibitory effect of PhAs-LHP. More importantly, PhAs-LHP inhibited PC growth in mice and enhanced the inhibitory effect of Gem on PC growth at 2 times lower molar concentration than PAO. These results indicate that PhAs-LHP inhibited PC more potently than ATO/PAO and suggest a potential clinical use for the combination of Gem with the peptide-linked arsenic compound for the treatment of pancreatic cancer.