Two simulations of pressurized water reactor (PWR) anticipated transient without scram (ATWS) sequences were performed in the loss-of-fluid test (LOFT) facility. These were designated as tests L9-3 and L9-4. Test L9-3 is a loss-of-feedwater transient, while L9-4 simulates an ATWS accompanied by loss of off-site power. In the latter case, the main steam valve and primary circulation pumps are tripped at the beginning of the experiment, along with the steam generator feedwater flow. The behavior of these experiments was analyzed with the RETRAN-02 computer code in order to evaluate the capability of the code to predict A TWS behavior against the experimental evidence exhibited by LOFT. The complex sequence of events, which occurs in L9-3, creates a difficult system modeling problem. The primary influence on the entire system response is the steam generator heat transfer rate (given core power as a boundary condition). Results of the analysis demonstrate how the course of the computer calculation is influenced by the steam generator model and its initial conditions. Test L9-4 does not contain the thermal-hydraulic complexity of test L9-3 due to the immediate isolation of the steam generator. Nonetheless, it reinforces the general conclusion that RETRAN-02 will produce an adequate simulation of PWR ATWS behavior if the initial and boundary conditions are completely defined.