From the removal of charcoal from the oven to its final use, several stages are involved, exposing the material to various atmospheric conditions, including variations in relative humidity. In the Amazon, climatic conditions such as high humidity and high temperatures aggravate this scenario, directly influencing the quality of charcoal produced in the region. The water content in charcoal intended for steelmaking varies from 5% in the dry season to 10% in the rainy season, reflecting the influence of climatic conditions on the quality of the bioreductant. This variation impacts the energy density and calorific value of the material, as well as the efficiency of steelmaking processes, which is crucial for optimizing energy consumption and reducing costs. This study investigates the hygroscopicity of charcoal produced from three Amazonian wood species Dinizia excelsa Ducke (angelim-vermelho), Inga spp. (inga) and Parkia pendula (faveira), carbonized at two final temperatures, 400°C and 800°C. The samples were conditioned under six controlled relative humidity levels ranging from 40% to 90% at 30°C. The analyses performed included basic density, immediate and elemental chemical composition, Higher Heating Value (HHV), surface area (BET), porosity, scanning electron microscopy (SEM), FTIR, thermogravimetric analysis (TGA), and equilibrium moisture determination. Near-infrared (NIR) spectral signatures were collected using a portable MicroNIR spectrometer, and multivariate chemometric models were developed to classify samples based on wood species and humidity conditions. Results showed that gravimetric yields were higher for Dinizia excelsa carbonized at 400°C (average 37%) and lower for Parkia pendula carbonized at 800°C (average 33%). The Higher Heating Value increased with carbonization temperature, reaching 33.86 MJ/kg for Dinizia excelsa at 800°C. Elemental analysis revealed an increase in carbon content with temperature, up to 93.52% for Dinizia excelsa at 800°C. Equilibrium moisture varied with temperature and species: charcoal carbonized at 400°C showed lower hygroscopicity (5.64% average for Dinizia excelsa), while charcoal at 800°C absorbed more moisture, with maximum values observed in Parkia pendula (20.60%). This increase in moisture content is likely related to structural changes in porosity caused by higher carbonization temperatures. Principal Component Analysis (PCA) indicated spectral variance associated with these factors, and Partial Least Squares Discriminant Analysis (PLS-DA) models achieved high classification accuracy for wood species (above 89%) and humidity conditions (above 94%). Cross-validation confirmed model robustness, with over 97% correct classification for relative humidity levels. This study demonstrates the feasibility of using portable NIR spectroscopy combined with chemometric modeling for rapid, non-destructive assessment of charcoal properties. Understanding the hygroscopic behavior of Amazonian charcoal contributes to improved production practices and more efficient use of these materials in the steel production chain, with industrial relevance especially in regions with high relative humidity and ambient temperature.