The objective of this study was to test the hypothesis that the roasting conditions used for hazelnuts, such as the air temperature, air velocity and roasting time (independent variables), could be optimized by using Response Surface Methodology. Effects of independent variables on sensory and physical characteristics were determined. A consumer test was used to determine the acceptable samples. Very dark and very light roasted samples, corresponding to 165 °C, 3 m/s, 25 min and 125 °C, 1 m/s, 15 min process conditions, respectively, were unacceptable. Superimposed contour plots were used to determine the values of independent variables and these showed the process conditions where all product characteristics were acceptable to consumers. At low velocity (0.3 m/s), acceptable products were produced at about 165–179 °C for 20–25 min. When air velocity increased, air temperature shifted to lower temperatures. Samples roasted at 145 °C, 2 m/s, 28 min, 165 °C, 1 m/s, 25 min and 145 °C, 3.7 m/s, 20 min produced the most acceptable products. The sample roasted at 165 °C, 1 m/s, 25 min required the least air velocity and was the most economical in terms of energy consumed among the samples rated most acceptable by consumers.