This study was an attempt to optimize the formulation of a shelf-stable fermented milk beverage using response surface methodology. The effect of whole milk powder-to-whey ratio, agar, starch, pectin, fructose and cream on the viscosity and water-holding capacity of the beverage was investigated. Whole milk powder-to-whey ratio, agar and starch were identified as the most significant factors using the Plackett-Burman design (P < 0.05). The levels of the four factors were optimized using response surface methodology with Box-Behnken design. A response surface model was developed indicating the relationship between each of the response variables and the four factors. The optimal combination obtained from the model was whole milk powder-to-whey ratio 5:1, pectin 1 g/L, starch 10 g/L and pectin 0.5 g/L.