The threshold stress, o, included in the creep equation to explain the high temperature behavior of discontinuously reinforced metal matrix composites (MMCs) is criticized on the basis of microstructural considerations and a new creep data analysis. An alternative interpretation, based on changes in the composite matrix microstructure and, in particular, a load transfer mechanism, is proposed. The resulting creep equation is similar to that in which o is used: in essence, o is simply replaced by the stress carried by the reinforcement (referred to as T). New creep data on 6061Al-15vol%SiCw composite and the corresponding un-reinforced alloy, allowing direct experimental assessment of composite creep strengthening, , are analyzed. The linear dependence found of with the applied stress, , ( correlates reasonably well with Shear-Lag and Eshelby models' predictions of T, transferred during composite creep deformation. The possible occurrence of damage mechanisms and the complexity of modeling these mechanisms to predict the overall composite creep behavior are also discussed.