Cobb-Douglas functional form has been widely used in estimating production function in traditional agriculture. However, Cobb-Douglas specification may not always represent the underlying nature of technology in different crop seasons. In this paper, the validity of the restrictions imposed by Cobb-Douglas functional form was tested vis-a-vis more general functional forms in a sample of small-holding farmers in Bangladesh. The test results showed that the Cobb-Douglas restrictions were validated against both transcendental and translog functions for aman rice, pulses, wheat, oilseeds and IRRI rice, but not in the case of aus rice and jute, the major competing crops in the Summer season. This, therefore, raises some doubts as to its ‘indiscriminate’ use in estimating disaggregated production functions in peasant agriculture. Testing of Cobb-Douglas ‘Myths’ : An Analysis with Disaggregated Functions in Bangladesh Agriculture.