Laboratory experiments and numerical simulations are performed to determine the accuracy and reproducibility of the falling-ball test for viscosity determination in Newtonian fluids. The results explore the wall and end effects of the containing cylinder and other possible sources that affect the accuracy and reproducibility of the falling ball tests. A formal error analysis of the falling-ball method, an evaluation of the relative merits of calibration and individual measurements, and an analysis of reproducibility in the falling-ball test are performed. Recommendations based on this study for improving both the accuracy and reproducibility of the falling-ball test are presented.