Accurate and mechanistically plausible mathematical models of DNA double-strand break (DSB) rejoining kinetics are needed to correctly estimate the dependence of cell death and transformation on linear energy transfer, radiation dose and time. When integrated into more comprehensive risk estimation approaches, such models are potentially valuable tools in applications such as treatment planning for radiotherapy. In this study, we compared 10 DSB rejoining models based on data collected from 61 mammalian cell lines after high-dose-rate photon or heavy ion irradiation. The set of models included formalisms with: 1. one, two or three discrete first-order rejoining rates; 2. continuously distributed first-order rejoining rates; and 3. second-order rejoining rates. The Akaike information criterion was used to quantify the relative support for each model from the data, accounting for goodness of fit and model complexity. The best performance was exhibited by a bi-exponential model with two discrete rejoining rates and a model with gamma-distribution rejoining rates. Models with more than three free parameters overfitted the data and models with single DSB rejoining rates or with an inflexible distribution of rejoining rates lacked accuracy. Of special note is that the analyzed data provide little support for models that rely on pairwise interactions to describe DSB rejoining kinetics. Consequently, kinetic cell survival models reflecting bi-exponential DSB rejoining might be preferable to models based on the kinetics of intra- and inter-lesion rejoining.