This paper compares the form of f(v||) arising from the warm-ion, kinetic models of Emmert et al. [ Phys. Fluids 23, 803 ( 1980) ] and Bissel and Johnson [ Phys. Fluids 30,779 ( 1987) ] with experimentally measured distributions from the DITE tokamak obtained with a retarding field analyzer (RFA). The results show that the commonly adopted assumption of a distribution accelerated through the sheath potential and approximating to a Maxwellian in the high-energy tail is reasonable. In this respect, Emmert’s distribution is more appropriate. In addition, the common practice of neglecting the presheath effect on the distribution is generally justified in analyzing the RFA characteristic, but may incur large error under some circumstances. While the comparison also illustrates how differing assumptions regarding the source function can have a reasonably strong influence on the ion velocity distribution at walls and limiters, the integrated distributions required for comparison with RFA data are sufficiently similar that distinguishing between the models experimentally is not generally possible. The problems of resolving these differences experimentally are discussed and suggestions are proposed for further experiments.