RNA interference (RNAi) is an exciting new tool to effect acute in vivo knockdown of genes for pharmacological target validation. Testing the application of this technology to metabolic disease targets, three RNAi delivery methods were compared in two frequently utilized preclinical models of obesity and diabetes, the diet-induced obese (DIO) and B6.V-Lep/J (ob/ob) mouse. Intraperitoneal (i.p.) and high pressure hydrodynamic intravenous (i.v.) administration of naked siRNA, and low pressure i.v. administration of shRNA-expressing adenovirus were assessed for both safety and gene knockdown efficacy using constructs targeting cJun N-terminal kinase 1 (JNK1). Hydrodynamic delivery of siRNA lowered liver JNK1 protein levels 40% in DIO mice, but was accompanied by iatrogenic liver damage. The ob/ob model proved even more intolerant of this technique, with hydrodynamic delivery resulting in severe liver damage and death of most animals. While well-tolerated, i.p. injections of siRNA in DIO mice did not result in any knockdown or phenotypic changes in the mice. On the other hand, i.v. injected adenovirus expressing shRNA potently reduced expression of JNK1 in vivo by 95% without liver toxicity. In conclusion, i.p. and hydrodynamic injections of siRNA were ineffective and/or inappropriate for in vivo gene targeting in DIO and ob/ob mice, while adenovirus-mediated delivery of shRNA provided a relatively benign and effective method for exploring liver target silencing.