Reducing lipid accumulation in insulin target tissues is critical for the treatment of type 2 diabetes. This study aimed to develop a biochemical assay in cells for high throughput (HTP) screening of anti-diabetic drugs by reducing lipid accumulation via different mechanisms. We designed a new method to extract triglyceride (TG) with KOH to allow biochemical quantification of TGs for HTP screening in 3T3-L1 cells. This new method was validated for its biochemical properties with identical results of TG obtained with or without KOH (r(2) = 0.9978, p < 0.001) and a fourfold improvement in TG extraction recovery rate (88-95%, p < 0.001) as compared to the conventional chloroform/methanol extraction (12-18%). The ability of this phenotype screening to capture potential anti-diabetic drugs was verified by pharmacological agents well known to alter lipid accumulation by different mechanisms including AMPK activators, fatty acid synthesis inhibitors, PPARγ activator and several lipogenic substrates. To further demonstrate the application of this screening tool for discovery of new anti-diabetic drugs, we screened >200 new candidates selected from Chinese medicine and identified 49 compounds from different classes which reduced TG content by >50% at 1 μM or >75% at 10 μM. Finally, we tested two selected leads (albiflorin and oxymatrine) in vivo and confirmed their efficacy in reducing visceral adiposity, glucose intolerance and hepatic steatosis in high fat-fed or high fructose-fed mice. Our results indicate that screening for the efficacy on lipid accumulation in cells by biochemical quantification of TGs with KOH extraction is an effective tool for the identification of new anti-diabetic compounds.