Having fast computers with lots of memory makes it possible to make performance predictions using mechanics, which simply could not have been done even as recently as 3 or 4 years ago. This trend can be expected to continue. Pavement materials, whether asphalt concrete, base course, or subgrade, can now be characterized in realistic ways that were simply unavailable to pavement engineers. The dependence of these materials on stress state, moisture, temperature, strain rate, and damage is what has made their characterization difficult, if not impossible. This complexity has posed an almost insuperable problem for computational mechanics and for the linkage of these properties to construction specifications on the one hand and performance of in-service pavements on the other. Although these tasks will never be simple, the tools to work on them are becoming more adequate. Some examples are given of how one can proceed to simplicity through complexity in several areas that are important to pavement performance. In every case, one must be able to measure a property of the material that can be used as input to a computer program, principally a finite element program, that allows material properties to change from point to point. Several examples are given in the categories of material properties, material behavior, materials testing in the laboratory, nondestructive testing in the field, and performance prediction models. Laboratory testing, materials characterization and properties, nondestructive testing of pavements in service, construction specifications, pavement variability and reliability, field data collection, performance prediction modeling, and prediction of pavement response and distress with modern computer methodstaken individually, these subjects look complex. Combined with the glue of mechanics, they are simplicity itself. And they confirm Alfred North Whiteheads principle, The only simplicity that can be trusted is that which lies beyond complexity.