Maintaining a dry bedding is key in dairy cattle management. High bedding moisture has been related to poor cows’ hygiene, increased risk of mastitis and reduced lying time. Aim of this study was to study water evaporation dynamics in a bedded pack for dairy cows. A custom method was developed to measure drying rate (DR) of different bedding materials contaminated with livestock excreta. The method involved exposing samples of bedding with defined volume (80 mL) and surface area (19.7 cm²) to different artificially controlled conditions. Four bedding materials were included in the study (sawdust, wheat straw, wood pellets and straw pellets). For each material, bedding was collected in a real farm, dried at 65°C, and then added with a controlled amount of water to achieve 2 levels of initial moisture (40%, 70%). Experiment was carried out in a climatic chamber, which allowed recreating 2 levels of air temperature (10°C, 25°C) and 2 levels of relative humidity (40%, 80%). A custom wind tunnel was constructed to simulate 2 levels of air velocity (0.0, 1.0 m/s). Also, to reproduce the heat developed by bacterial activity in a real bedded pack, the custom wind tunnel was provided with a temperature-controlled heating plate, which allowed exposing sample to 2 levels of heating (ambient vs 35°C). A crossed treatment design with 2 repetitions was used resulting in 256 samples tested in total. Drying rate was calculated as the difference in weight before and after a 24h exposure to the test conditions. Data were analyzed using Random Forest regression. Forest Floor method was employed to visualize and interpret the model obtained. Average DR across all treatments was 0.56 kgH₂O/m²*day, ranging from 0.08 to 1.67 kgH₂O/m²*day. Variable importance analysis showed that initial sample moisture, air velocity and relative humidity were the most important variables affecting DR. Both air temperature and sample heating showed to produce a relatively small effect on evaporation. Also, DR resulted to be scarcely related to type of bedding material, which resulted to be the least important variable.