The prevailing state of knowledge of two-phase flow in complex systems, as in the nuclear field, usually leads a certain degree of ad hoc calibration of computational models and the consequent inevitable subjectivity in analyses and assessments. The UPC-ANT Uncertainty Analysis methodology for code assessment attempts to maintain scientifically desired features of models and analyses, by crediting BEPU analysis and centering the assessment criteria on T-H phenomenology rather than on safety criteria, as it is the usual practice. An SB-LOCA in the upper head is utilized as a case study, and global, chronological, FOMs and phenomenological analyses are carried out and illustrated briefly. Overall, it was concluded that the developed RELAP5 model reproduces all the relevant T-H phenomena to the investigated scenario, either partially or totally, and the BEPU methodology is generally adequate to perturb the underlying T-H phenomena.