In this paper, we study the return to human capital variables for wages of workers observed in Tunisian matched worker-firm data in 1999. This reveals us how returns to human capital in a Less Developed Country like Tunisia may differ from the industrial countries usually studied with matched data. We develop a new method based on multivariate analysis of firm characteristics, which allows us most of the benefits obtained by introducing firm dummies in wage equations for studying the effect of education. It also provides a human capital interpretation of the effect of these dummy variables. Moreover, in the studied data, using three firm characteristics easily collectable yields results close to those obtained by using the matched structure of the data. The workers with low wages or low conditional wages experience greater returns to human capital than workers belonging to the middle of the wage distribution, while their return to schooling is significantly lower than that of high wage workers. The estimates support the hypothesis that human capital is associated with positive intra-firm externality on wages. Therefore, a given worker would be more productive and better paid in an environment strongly endowed in human capital. However, the low wage workers do not take advantage of the human capital in the firm. Conversely, the low wage workers benefit from working in the textile sector in terms of wages unlike the middle and high wage workers. Finally, the low wage workers and high wage workers benefit from an innovative environment, while the middle wage workers do not.