Wildland-Urban Interface (WUI) and Wildland fires cause damages to infrastructure and the environment, resulting in loss of life, property, and an increasing cost of mitigation. Firebrands are known as one of the leading fire spread mechanisms in wildland and WUI fires. A more thorough understanding of the firebrand phenomenon and its role in wildfire spread can lead to the development of better predictive models as well as developing effective mitigation strategies for communities prone to these fires.Firebrand characterization studies to date have been limited to reporting the mean values of physical properties of firebrands based on a small number of samples without a statistical framework. This study presents an effort toward building a statistics-based structure and developing mathematical models to characterize the physical properties of firebrands from selected vegetative fuels. Full-scale firebrand production experiments were designed and carried out in a large wind tunnel facility. Firebrand collection, characterization, and data analysis were based on the statistics-based framework. Firebrand properties (including mass, projected area, and flying distance) were measured and characterized by vegetation species and wind speed. The outcome of these data was used to find the best probability distribution function and potential correlations among random variables. Results of statistical tests indicated that the likelihood of fitting a lognormal distribution over this dataset was the highest, and this probability density function was the best fit for this dataset Afterward, a regression model was selected and fitted based on the results of a goodness-of-fit statistical test. A correlation study in vegetation type and species level showed a strong relationship between the mass and projected area of firebrands regardless of fuel characteristics (i.e., vegetation type and species level) and environmental conditions of the experiment. It also indicated that higher wind speeds are capable of generating firebrands with greater weight and larger projected area. This is an essential takeaway as the accumulation of firebrands on a recipient fuel may start a spot fire, depending on numerous parameters of landed firebrands, receptivity of the fuel, and environmental factors. Nevertheless, based on both factors, it can be concluded that firebrands generated in higher wind speeds are potentially more lethal. Statistical tests in this study showed that higher wind speeds affect the firebrand generation in all vegetation types, although the magnitude of the effect differs by vegetation species. Statistical comparison of the physical property values of firebrands showed that more massive firebrands were generated in higher wind speeds as well as a larger number of generated firebrands. Results also showed that with 95% confidence, the mass values of firebrands were between 0.02 and 0.33-gr, and values for the projected area were between 0.71 and 2.49-cm^2. Methods developed in this study brought a better understanding of the firebrand phenomenon (particularly firebrand generation). Interpretation of these findings can help various stakeholders in wildland and WUI fire protection in revising the wildfire mitigation strategies. Examples include property owners, the fire/materials and standards engineering communities, government agencies, and the fire and emergency services.