In this retrospective cohort study at a quaternary referral hospital, 829 adult patients supported by extracorporeal membrane oxygenation (ECMO) between January 1, 2014, and October 31, 2024, were screened for ventilator-associated pneumonia (VAP). The incidence of VAP in ECMO-supported patients was 17.7 per 1,000 ECMO days (venoarterial [VA] ECMO, 22.1 per 1,000 ECMO days; venovenous [VV] ECMO, 13.8 per 1,000 ECMO days). Median time from ECMO cannulation to diagnosis of VAP was 4 (interquartile range [IQR], 3–8; range 1–84) days. Gram-negative bacteria comprised most respiratory pathogens. A significant increase in minimum ECMO sweep was noted from 2 days before to 1 day after VAP diagnosis (0.79 L/min increase, p = 0.004). Only 38 of 153 patients (24.8%) met National Healthcare Safety Network (NHSN) VAP definitions. After creating a modified VAP definition, which included an ECMO sweep increase of 0.5 L/min as evidence of worsening respiratory status, 78 (51.0%) met the modified definition. Those who met the modified VAP definition had worse in-hospital (60.3% vs. 28.0%, p < 0.001), 30 day (33.3% vs. 17.3%, p = 0.023), and 90 day (53.8% vs. 25.3%, p < 0.001) mortality compared with those who did not meet the modified VAP definition. The incorporation of ECMO parameters into diagnostic criteria may improve detection sensitivity, but further validation is necessary before implementation.