Hospital fire safety standards in Saudi Arabia remain primarily prescriptive and do not explicitly account for patient mobility constraints, complex building geometry, and time-dependent fire dynamics. The 2015 Jazan Hospital fire, which claimed 25 lives and injured more than 100 people, demonstrated that code compliance alone cannot guarantee occupant safety. This study presents a four-phase BIM framework integrating Autodesk Revit, PyroSim/FDS fire simulation, and Pathfinder agent-based evacuation simulation, applied to a 250-bed hospital in Qatif, Saudi Arabia. The FDS computational mesh was assessed at D*/δx = 12.2 against NIST NCSTAR 1-5E benchmark data (mean errors: temperature 6.8%, smoke optical density 7.2%), and Available Safe Egress Time (ASET) was determined using the multi-criterion ISO 13571:2012 methodology. Under the examined SBC 2018 baseline (Scenario 1, four corridor-end exits), smoke penetrated the ICU corridor within 60 seconds, yielding ASET = 118 s and a Monte Carlo Required Safe Egress Time (RSET) of 412 ± 35 s (n = 50; approximate 95% CI for the mean: 402–422 s), an ASET–RSET deficit of 294 seconds. Scenario 2, incorporating fire-rated ICU/OR compartmentation and ten distributed emergency exits, extended the reported ASET beyond 185 s and reduced mean RSET to 281 ± 22 s (approximate 95% CI for the mean: 275–287 s), a descriptive reduction of 32%. On the reported ASET range of 185–225 s, however, RSET remains greater than ASET; the residual deficit is 56–96 s. Sensitivity analysis across three ignition locations indicates that the Scenario 1 deficit is not specific to one fire origin. The findings support performance-based BIM simulation as a complement to prescriptive code assessment in hospitals.