The proton temperature ratio T⊥/T ∥ measured by the AMPTE/CCE spacecraft under conditions of high solar wind dynamic pressure follows closely the relation T⊥/T∥ =(T⊥/T∥)&bgr;≡1+0.85&bgr;-0.48∥. On the basis of this relation, a bounded anisotropy fluid model was developed which describes the temperature evolution of a collisionless anisotropic (T⊥≠T∥) plasma including the effect of energy exchange due to pitch angel scattering by cyclotron waves. This model has been shown to well predict the ion temperature evolution within the magnetosheath close to the magnetopause as observed by CCE. The model was also successful in explaining the temperature evolution of data from the AMPTE/IRM spacecraft averaged over more general solar wind conditions, despite the fact that much of the IRM data lay well below the CCE relation. Here, we make a more thorough comparison of the bounded anisotropy model with the IRM data. Three example IRM crossings are shown: (1) a case with the temperature ratio T⊥/T∥ closely following the CCE relation, (2) a case with T⊥/T∥ closely following the functional form of the CCE relation, but at a lower value, and (3) a case where T⊥/T∥ often falls well below and evolves in a manner unrelated to the CCE relation. These three cases roughly exhibit the range of typical behavior in the entire IRM data set. In all three cases, it is clear that the CCE relation represents an approximate upper bound on the anisotropy and that the ion cyclotron waves control the temperature evolution when the anisotropy is driven up to this bound. The bounded anisotropy model is successful in explaining the temperature evolution in the first two cases and the gross features of the third. In the third case, large fluctuations in temperature were probably due to temporal changes in magnetosheath conditions. ¿ American Geophysical Union 1995