Abstract In reliability analysis of technical systems the failure frequency of cut sets alone is often insufficient to assess the probability of an undesired event. In many systems an undesired event occurs after failures of system elements, however not before a limited time period is over. Since system elements can be repaired the delay times can have a crucial effect on the probability of an undesired event. Here a method is developed to consider as many different delay times as desired in calculating the probability of the fault tree top event.