Two mathematical models, power-law (Kovatch et al, J Dent Res 1976;55:783-786) and Maxwell-Weichert (Stevenson and Kusy, Angle Orthod 1994;64:455-467), have been proposed to describe the force decay of polyurethane orthodontic elastomeric chains. Purpose: Perform carefully controlled in vitro force degradation measurements on a variety of commercial elastomeric chain products to determine which mathematical model yields the better fit to the observed data. Methods: Specimens of 13 products, extended to 25 mm displacement, were tested as-received and after 24 hr presoaking in a 37°C artificial saliva bath. Measurements for specimens strained continuously for 6 hr with a factory-calibrated digital force gauge were compared with measurements obtained from specimens transferred to the gauge after aging on a stretching rack similar to that used in many previously published studies. For all of these measurements, force data taken at equally-spaced logarithmic time points, and normalized to induced strain and initial force, were used to represent degradation curves for individual products. Nonlinear regression analyses were used to fit both mathematical models to each product and conditioning state. Results: A Welch ANOVA comparing the equivalence of force at 19.2 hr showed that the force gauge measurements were significantly higher (p < 0.001) than those obtained with the stretching rack. Therefore, only force gauge data were included in the subsequent nonlinear regression. Three-factor repeated-measures ANOVA revealed that the mathematical model used to fit the force values as a function of time significantly affected the residual error, which was significantly lower (p < 0.001) for power-law, compared with Maxwell-Weichert, predictions. Conclusions: The power-law model provided a more accurate prediction of observed force at all time points for all products. Results do not validate the stretching rack and intermittent force measurement methodology used in previous investigations.