This research empirically investigates the well known 'poor-but-efficient' hypothesis formulated by Schultz (1964) assuming that small scale farmers in developing countries are reasonably efficient in allocating their scarce resources by responding positively to price incentives. Deviating from Schultz it is assumed here that scale effects explain a considerable proportion of small scale farmers' relative efficiency. The theoretical underpinnings of the scale efficiency concept are briefly reviewed before a normalized generalized Leontief profit function is modeled by using its output supply and input demand system to capture the joint production of cassava flour and maize by a sample of small scale farmers in the Bragantina region of the Eastern Amazon, Brazil. The discussion on theoretical consistency and functional flexibility is considered by imposing convexity on the GL profit framework. The empirical results confirm our revised hypothesis that small farmers in traditional development settings are 'poor-but-allocatively efficient' by clearly suggesting considerable inefficiency with respect to the scale of operations.