Pavements are complex physical structures responding to the influence of numerous environmental, subsurface, and load-related variables and their interactions. Subsequently, the task of predicting the multi-faceted responses of pavements to the series of interrelated variables is complex and must be addressed by using a number of assumptions and simplifications. This paper presents a methodology used to develop deterioration models for the primary highway network in the province of Alberta, Canada. A major assumption is that a road is considered to be deteriorated and reached its service life limit when its roughness reaches a specified trigger value. The international roughness index (IRI) was taken as the roughness measurement in this research, with a value of IRI equal to 2.8 considered the trigger for initiating rehabilitation action. Approximately 1700 road segments comprise the primary highway network in Alberta. Each of these segments has its own set of attributes. A stratification methodology was used to classify and group these 1700 segments into eight groups, each possessing the same characteristics in terms of its life cycle.