Historically, Leyte Gulf in central eastern Philippines has received catastrophic damage due to storm surges, the most recent of which was during Typhoon Haiyan in 2013. A city-level risk assessment was performed on Leyte Gulf through synthetic storm generation, high-resolution ocean modeling, and decision tree analyses. Cyclones were generated through a combination of a Poisson point process and Monte Carlo simulations. Wind and pressure fields generated from the cyclones were used in a storm surge model of Leyte Gulf developed on Delft3D. The output of these simulations was a synthetic record of extreme sea level events, which were used to estimate maximum surge heights for different return periods and to characterize surge-producing storm characteristics using decision tree analyses. The results showed that the area most prone to surges is the Tacloban–Basey area with a 2.8 ± 0.3 m surge occurring at a frequency of every 50 years. Nearby Palo area will likely receive a surge of 1.9 ± 0.4 m every 50 years while Giporlos–Salcedo area a surge of 1.0 ± 0.1 m. The decision tree analysis performed for each of these areas showed that for surges of 3–4 m, high-velocity winds (> 30 m/s) are consistently the main determining factor. For the areas, Tacloban, Basey, and Giporlos–Salcedo, wind speed was also the main determining factor for surge > 4 m.