Due the different possibilities for fit the Probability Density Functions adjustable to a wind speed data set, a best fit selection criterion is developed based on slope, intercept values and standard errors of Ordinary Linear Regression model calculated from the probabilistic model and experimental data. Uncertainty associated with measuring instruments is analyzed, and an interpretation is presented in terms of the electric power generated. In addition, a methodology is proposed to generate scenarios of energy production used in financial evaluations, which is possible since the wind speed data used retain its uncertainty. The relevant conclusions are that a sampling technique based on representative average wind speeds does not reproduce the original distribution of wind speed data set, since for the observed sample, the parameters of the fitted distributions vary depending on sampling time. Accordingly, assessments based on this sample technique leads to a resource underestimation. (C) 2012 Elsevier Ltd. All rights reserved.