Understanding the resuspension of droplets from surfaces into air is important for elucidating a range of processes such as disease transmission of airborne pathogens and determining environmental contamination and the effectiveness of cleaning procedures. The resuspension condition is defined as the escape velocity of a droplet from a surface. This study investigated the dynamics of microliter-sized droplet resuspension off surfaces utilizing a novel free-fall device. We studied surfaces with three different wettabilities, three droplet volumes, and substrate velocities ranging from 0 to 3.5 m/s for deionized water and viscous droplets representing a prototype saliva substitute. Experimental results provide quantitative results for the increased propensity for drop resuspension for more hydrophobic surfaces, larger droplet volume, and higher velocity. By using high-speed imaging, we segment the resuspension process into four stages: initial equilibrium, deformation, elongation, and breakage. Experimental results are generalized as a machine-learning-derived decision surface, which predicts resuspension by defining a 2D decision boundary in our 3D parameter space. We present a simple physical model, corroborated by computational fluid dynamics simulations, for the dynamics of resuspension that explains the process and is in good agreement with the experiments.