The need for demand forecasts is common in the planning and control processes. Such predictions represent a key factor for managing the various organizational departments, especially those related to logistics and supply chain such as stock management, service level and purchasing. In this context, this study s objective was to propose improvements in Brazilian Navy s health material through the implementation of more effective demand forecasting methods. In the Brazilian Navy, the demand data series for essential health items have an erratic pattern, making the mathematical modeling for demand forecasting a complex task. The methods tested and evaluated are simple, practical, and low-cost deployment. Are they: Exponential Smoothing, Croston method and Poisson Distribution, besides the current demand forecasting currently applied by the Brazilian Navy itself. The results found in this dissertation showed a 40 percent improvement in the main items demand forecast when compared to the current method, It s implementation will provide significant savings of financial resources and increase the level of service of health items This work serves as a basis for use in other supply chains in Brazilian Navy (spare parts, fuels, ammunition, foodstuffs and uniforms), and can be implemented in both MB and other Armed Forces.