This study presents a comprehensive experimental and probabilistic analysis of wind-driven building-to-building fire spread, focusing on the interplay between exposure intensity, duration, and material response. Twenty-three full-scale tests were conducted under varied wind speeds, separation distances, and material configurations to capture a spectrum of damage levels and move beyond the conventional binary outcomes of most fire experiments. As fire intensity builds, the target structure experiences increasing heat exposure, while collapse of the source structure during the growth phase disrupts flame dynamics and causes abrupt intensity drops. This deviation from the classic growth, steady, and decay pattern weakens the correlation between observed damage and traditional metrics such as peak heat flux or long-duration heat load. To address this, damage classifications from cosmetic to destroyed were mapped to statistical distributions of energy fluence accumulated over a characteristic intermediate timescale, better reflecting material response under realistic fire conditions. The resulting probabilistic framework supports risk-informed decisions aligned with the acceptable level of risk. At a separation of 10 ft (3 m), the tested building materials showed minimal likelihood of survival when exposed to fires from large, fully loaded sheds. By about 20 ft (6 m), the fire exposure decreases to a level where resilient building components have a fighting chance, and survival probabilities improve substantially at 30 ft (9 m). This framework provides a statistical view based on a physics-based interpretation of ignition thresholds to quantify intermediate damage and assess vulnerability under variable fire exposures, supporting fire-resilient community design and mitigation.