RELAP5 (reactor excursion and leak analysis program, version 5) code analyses were performed on two ROSA/LSTF (rig of safety assessment/large scale test facility) experiments on PWR (pressurized water reactor) safety system that simulated cold leg small-break loss-of-coolant accidents with 8-in. or 4-in. diameter break using SG (steam generator) secondary-side depressurization. The SG depressurization was initiated by fully opening the depressurization valves in both SGs immediately after a safety injection signal. In the 8-in. break test, loop seal clearing occurred and then core uncovery and heatup took place by core boil-off. Core collapsed liquid level recovered after the initiation of accumulator coolant injection, and long-term core cooling was ensured by the actuation of low-pressure injection system. In the 4-in. break test, on the other hand, there was no core uncovery and heatup due to smaller break flow rate than in the 8-in. break test. Adjustment of Cd (break discharge coefficient) for two-phase discharge flow predicted the break flow rate reasonably well. The code well predicted the overall trend of the major thermal-hydraulic response observed in the two LSTF tests by the Cd adjustment. The code, however, overpredicted the peak cladding temperature because of underprediction of the core collapsed liquid level due to inadequate prediction of the accumulator flow rate in the 8-in. break case.