Rationale: Childhood pneumonia is a leading cause of death worldwide. Household air pollution (HAP) from biomass cooking fuelmay be an important child pneumonia risk factor. It is unknown whether liquefied petroleum gas (LPG) cookstoves reduce severepediatric pneumonia. Methods: We conducted a randomized controlled trial of an LPG cookstove, continuous fuel distribution andbehavioral messaging, compared to usual biomass-burning cookstove use, among 3,200 pregnant women and their householdsin 10 resource-poor settings in India, Guatemala, Peru, and Rwanda between May 2018-September 2021. Pregnant women 18-34 years old and 9-19 weeks gestation (by ultrasound) were eligible; follow-up was through the offspring’s first birthday. Wemeasured 24-hour fine particulate matter (PM2.5), black carbon and carbon monoxide at baseline, pregnancy, and during thechild’s first year of life. Cookstove use was monitored for compliance. Active surveillance for severe pneumonia in hospitals andselected health centers was conducted using the following case definition: observed and/or reported cough and/or difficultbreathing, and (1) >1 World Health Organization (WHO)-defined general danger sign with ultrasonographic or radiographicalveolar consolidation, or (2) hypoxemia; or (3) any pneumonia death by verbal autopsy. Secondary WHO pneumonia casedefinitions were also analyzed. All case episodes were included accounting for intra-child correlation. Analyses were intention-to-treat. Results: 3,200 pregnant women were randomized and there were 3,061 live births. LPG and control group antenatal PM2.5levels were 24.0 μg/m3 and 70.7 μg/m3, and antenatal traditional stove use in the LPG arm was a median of <1 day/month.Baseline characteristics of liveborn children were balanced between arms (Table). Overall, there were 175 severe pneumoniacases, 67% (n=118) in <6 month olds and 51% (n=89) in females. Hypoxemia was frequent (69%, 120/175) and eitherultrasonographic or radiographic consolidation occurred in 49% (86/175) of cases. We observed a severe pneumonia incidencerate ratio (IRR) of 0.96 (95% confidence interval, 0.70, 1.32; p-value=0.81) during the first year of life in the LPG arm compared tothe control arm. Twelve children died from pneumonia, 8.8% (n=8) were controls and 4.7% (n=4) were from LPG households.Although secondary WHO pneumonia cases were more frequent than primary definition cases, they had qualitatively similar IRRs.Conclusions: Overall pneumonia incidence and detection may have been affected by the COVID-19 pandemic. Despite PM2.5exposure decreases and high intervention compliance, we found no difference in severe pneumonia incidence among infants ofwomen randomized to a multi-component LPG intervention than usual biomass-burning cookstove practices.