Variations in the engineering demand parameter (EDP) of a region significantly affect the fragility curve, which is believed to impact the estimation of earthquake losses; hence, there is a need to adjust for different regional characteristics. In earthquake-prone regions that already have empirical EDP databases, fragility curve development generally uses this information. Regions without EDP databases require additional effort to start their development or adopt existing methods as a short-term solution. Some studies show that direct adoption is oblivious to the consequences of the resulting estimation deviations. This study investigates the effect of EDP variations derived from an analytical method—incremental dynamic analysis (IDA)—and two empirical methods—HAZUS and RISK-EU—on seismic loss estimation in typical school buildings in Bandung City, Indonesia. Three school buildings are used as case studies, with existing structural data collected through non-destructive testing on each building used for the analytical method. The observed earthquake losses are estimated in a single hazard scenario at eight return periods and in the form of annualized earthquake losses (AEL). The results of this study illustrate that the EDP variation has a significant impact on loss estimation based on the relative difference determined with the analytical method. The sensitivity analysis results indicate that the HAZUS method has a relative deviation of 2.61%–74.62% in the single hazard scenario and 19.66%–71.90% in the AEL, and the RISK-EU method shows a relative deviation of 3.48%–672.03% in the single hazard scenario and 53.44%–222.82% in the AEL. Simultaneously, the absolute deviation in the single hazard scenario shows that the HAZUS method has a deviation of <12%BRC (building replacement cost) and the RISK-EU method <15%BRC. The absolute deviation value can be utilized as a reference when considering directly adopting empirical methods in developing countries that do not have an EDP database.