The optimum protein extraction from fermented and non-fermented Perah seed (Elateriospermumtapos) was investigated using the response surface methodology (RSM). A box-behnken design with three independent variables which are NaOH concentration (6, 8 and 10%), extraction time (10, 20 and 30 minutes) and solvent/meal ratio (50:1, 100:1 and 150:1, v/w) was used to study the response of protein yield. A second-degree equation for independent and response variables was produced from simulation to obtain the contour plot graphs. The best protein extraction procedure from both fermented and non-fermented Perah seed was obtained at 5.5% of solvent extraction, 40:1 ratio of solvent/meal and at 32 minutes of reaction time. Surface concentration and ratio of solvent/meal were found to influence the protein yield from fermented seed; meanwhile only the solvent concentration influenced protein yield from non-fermented seed. The maximum protein yields for both fermented and non-fermented Perah seeds were 18.0 g/100g and 5.0 g/100g seed meal respectively.