Physics-based approach to the cryogenic flow health management is presented. It is based on fast and time-accurate physics models of the cryogenic flow in the transfer line. We discuss main features of one of these models - the homogeneous moving front model - and presents results of its validation. The main steps of the approach including fault detection, identification, and evaluation are discussed. A few examples of faults are presented. It is shown that dynamic features of the faults naturally form a number of ambiguity groups. A D-matrix approach to optimized identification of these faults is briefly outlined. An example of discerning and evaluating faults within one ambiguity group using optimization algorithm is considered in more details. An application of this approach to the Integrated Health Management of cryogenic loading is discussed.