The estimation of models of activity choice in LDCs (Less Developed Countries) is complicated by specific features of labour markets. In particular, entry into activity sectors is often restricted and worked hours are rationed. Moreover, in the informal sector workers may obtain their work rewards directly through technology with decreasing returns to labour. In this situation, complex structural modelling is often reputed infeasible. In this paper, we show that this statement is exaggerated and that it is possible to simultaneously estimate preferences, informal technology and sector entry costs. We estimate a structural non-separable model of activity choice and labour supply of female workers for formal, informal and inactivity sectors in the capital city of Cameroon. Our results provide evidence for the concavity in hours of the earnings function in the informal sector. We also estimate significant effects for several explanatory variables of the preferences and the earnings function. This allows us to contrast the labour supply elasticities for different sectors in a structural framework. However, estimation using a small sample is difficult and requires strong restrictions. In particular, there exists a trade-off between incorporating many explanatory variables in the model, and accounting for the non-linearity and the consistency of the structural explanation. These difficulties raise questions about the interpretation of simple models that are not simultaneously based on preferences, technology and entry costs.