A methodology was developed to calculate the time-dependent system failure probabilities of a pressurized steel pipeline segment containing multiple active metal-loss corrosion defects. The methodology models the pipeline segment as a series system and distinguishes three distinctive failure modes, namely small leak, large leak and rupture. The limit state functions associated with these failure modes are established based on typical industry practice as well as the burst and rupture prediction models available in the literature. The pipeline internal pressure is characterized as a simple stochastic process consisting of a sequence of independent identically distributed random variables each acting over a period of one year. The magnitude of a given sequence follows the annual maximum pressure distribution. A pipeline example, assumed to contain multiple identical corrosion defects detected and sized by a recently run inline inspection, was used to illustrate the methodology and to investigate the impact of the spatial variability of the pressure loading and pipe resistances at individual defects on the system failure probability. It was observed that the spatial correlation between the pipe properties (i.e. wall thickness and tensile strength) associated with different defects has a negligible impact on the system failure probabilities. The impact of the spatial correlation between the initial sizes of different defects on the system reliability was found to be dependent on the initial defect sizes; that is, the spatial variability of large initial defect sizes has a greater impact on the system reliability than that of relatively small initial defect sizes. The assumption of independent pressure loading at individual defects has virtually no impact on the probability of small leak; this assumption results in overestimated probabilities of large leak and rupture. The degree of overestimation can be significant for the probability of rupture if the number of defects contained in the pipeline segment is relatively large, e.g. greater than 10. The spatial correlation between the growth rates associated with individual defects was found to have a marked impact on the system failure probabilities, especially the probability of small leak. The probability of small leak evaluated based on independent growth rates can be more than six times as high as that corresponding to fully correlated growth rates. Furthermore, the probabilities of small leak and rupture decrease in an almost linear fashion as the coefficient of correlation between the growth rates associated with individual defects increases from 0 (independent) to 1 (fully correlated). On the other hand, the probability of large leak decreases more rapidly as the coefficient of correlation between the growth rates increases.