Abstract Although fiber reinforced polymer composites are increasingly used through the wet layup process for the rehabilitation of deteriorating and understrength concrete structures, there is very little validated information regarding performance over extended time periods. This has resulted in the use of excessively conservative factors in design in some cases, and unconservative estimates, or complete disregard of degradation effects on some characteristics, in others. In this paper two commonly used predictive approaches are used to provide estimates of long-term deterioration for a range of material characteristics which are compared to experimental data obtained over a 3 years period of exposure. It is shown that although predicted durability of tensile characteristics is extremely good for thin sections used conventionally, the rates of deterioration increase significantly with increase in the number of reinforcing layers used. In addition there is deterioration of other characteristics, especially related to interlaminar and intralaminar properties, that need to be considered. It is noted that both methods are unable to accurately account for effects of initial post-cure seen in ambient cured carbon/epoxy systems, although the conventional Arrhenius predictions provide good correlation with experiments once this mechanism has ceased. A methodology that is capable of accounting for temperature variation during exposure is also outlined. The predictions are compared with safety factors prescribed by ACI-440 and TR-55 and some inherent disadvantages of such approaches is highlighted. The basis for the estimation of safety factors based on a reliability approach using Weibull parameters is presented and typical results are shown, emphasizing some weaknesses in current design methodologies in the area of FRP rehabilitation of concrete.