This paper considers the apparent conflict between two types of policy conclusions regarding urban job creation as a response to the urban unemployment problem: (i) because of the Todaro paradox, job creation will lead to increased unemployment, and hence is not a useful policy; (ii) a subsidy for the employment of manufacturing labour (as shown by Harris and Todaro) is welfare-improving even in the presence of urban unemployment. It is argued that these conclusions are based on fundamentally different views of the rural-urban migration process in the two types of models, and a synthesis is proposed.