The American LTPP program has one of the greatest amounts of stored information regarding pavement characteristics and the evolution of their condition on a network. This is a valuable tool to support decision-making in what concerns maintenance and rehabilitation of pavements carried out by highway infrastructure agencies. This paper presents a study carried out to analyze the evolution of pavement conditions of several SPS-5 test sections, which were included in pavement rehabilitation studies. The evaluation considers approximately 15 years of data records, in California, Texas and Mississippi. In what concerns the development of pavement distresses, the evolution of longitudinal cracking, transverse cracking, and fatigue cracking (alligator cracking), International Roughness Index (IRI) and rut depth were analyzed. The evolution of pavement conditions represented by indicators related to the analyzed pathologies was considered as a function of several variables linked to rehabilitation techniques, such as preparation of the working surface (milling before construction of a new layer), thickness of overlay (50 or 125 mm) and the type of mixture applied (a new mixture or a mixture incorporating about 30% of recycled material). For the situations in which data were consistent and in sufficient quantity, multivariate regression analysis are presented in order to predict the evolution of pavement conditions with time, taking into account a set of explanatory variables related to the rehabilitation techniques applied and the traffic on pavements. The aggregate analysis of data for each state allowed us to conclude that preparation of the surface by milling, increasing thickness of overlay and the use of new mixtures or recycled asphalt interfere with the evolution of degradation. The influence of these factors varies for different types of pathologies under analysis.