Cattle body temperatures were measured under semi-controlled hot cyclic chamber temperatures. The four-parameter nonlinear PET model, is used to estimate body temperature in cattle challenged by heat stress. For each steer, the parameters can be estimated by the Nlin procedure and the sensitivity of each animal can be studied. It is also desirable to generalize the results by using the Nlmixed procedure to combine both the fixed and random effects. When comparing the results from the two procedures, we found heterogeneity among animals and/or days caused convergence problems for proc Nlmixed. Simulation studies were used to study how deviations from homogeneity effected the accuracy of parameter estimates, coverage of confidence intervals, and measures of nonlinear behavior when using the PET model to describe the dynamics of heat stress in cattle.