To date, empirical investigations of trade liberalization, under the conditions of increasing returns to scale (IRS) and imperfect competition (IC), have either assumed or imposed the market and productive structures necessary for such a model. However, of the recent IRS/IC models used to simulate the effects of trade liberalization, none have empirically tested for the presence of increasing return to scale prior to the analysis. With Tunisian data (1971-2004) and rigorous test procedures, we investigate evidence of IRS at the industry level. Using an econometric approach based on the estimation of the translog cost function and its associated cost share equations, we identify the sectors characterized by increasing returns to scale. Analysis of the results shows that specification of the model is sensitive to inclusion of time trend representing technology. For most sectors, the model accounting for technology did not fit the data well.Estimation results without time trend interactions are different and most sectors show signs of increasing returns to scale.