Determination of the probability distribution function for the time to failure is essential for the development of pavement life models, because the probability distribution function reflects the variability in pavement degradation. The pavement life and failure time are associated with the number of equivalent standard axle load applications for which the degradations reach a critical level. When the critical degradation level is reached, maintenance and rehabilitation work needs to be done to improve pavement condition. Research was undertaken to identify the appropriate statistical models for determination of the probability distribution function for the time to failure of pavement structures. The study used the rutting data collected on a test lane at the first full-scale accelerated pavement test in Louisiana. The research indicated that closed-form solutions or Monte Carlo algorithms can be used when the degradation models have a known form. The bootstrap algorithm can be used to determine the confidence intervals for probability of failure at a given time. If the form of the degradation model is not known, the survival analysis method based on censored observations must be used. The methods can be used not only for rutting life models but also for other pavement life models: cracking initiation time, cracking life, roughness, and serviceability lives.