Optical system designers need to estimate the expected PMD coefficient of the links in order to calculate the corresponding power penalty. Fiber PMD values varies randomly, therefore the link PMD estimation shall be calculated by statistical methodologies. This thesis presents a systematic study of PMD estimation of the amplified spans of two backbones. Three statistical techniques: Monte Carlo Technique, Gamma Model and Generalized Central Limit Theorem, recommended by Method 1 from TIA/EIA TSB 107 Guideline were used and the results compared with the field measurement values of these spans. Evaluation results showed that this methodology cause an overestimation of the link PMD coefficient due to a limitation of the measurement eqquipments and consequently an overestimation of the power penalty. This overestimation is due to the fact that fiber PMD is bellow of the resolution limit of the commercially available measurement equipments. Link PMD coefficient estimation needs the use of more precise measurement equipment (Not available yet) or to correct this calculus methodology for the optical link PMD estimation with the use of longer fibers/cables.