We report on estimating probability of failure (PoF) of an FA/18 bulkhead based on empirically obtained and published in open literature fatigue-crack growth and equivalent pre-crack size (EPS) distribution data. We demonstrate that the tail of the EPS distribution has a significant effect on PoF. Considering each flight to be a successful (i.e., no failure) fatigue test, we use a Bayesian approach to obtain an updated (posterior) EPS distribution and provide more accurate estimates of PoF. We then consider monitoring fatigue cracks in high stress concentration areas using fatigue damage sensors and show that using such sensor data (notably "no crack found" response) leads to a significant reduction of uncertainty in estimating the PoF. We also show the effect of increasing sensor accuracy (i.e., reliable detection of smaller cracks) on PoF predictions and required sensor interrogation intervals. The reported approach allows us to perform tradeoff studies on sensor accuracy and interrogation frequency for maintaining required levels of PoF.