Petrol stations emit benzene and other contaminants that have been associated with an increased risk of childhood leukemia. We carried out a population-based case-control study in two provinces in Northern Italy. We enrolled 182 cases of childhood leukemia diagnosed during 1998-2019 and 726 age- and sex-matched population controls. We geocoded the addresses of child residences and 790 petrol stations located in the study area. We estimated leukemia risk according to distance from petrol stations within a 1000 m buffer and amount of supplied fuel within a buffer of 250 m from the child's residence. We used conditional logistic regression models to approximate risk ratios (RRs) and 95% confidence intervals (CIs) for associations of interest, adjusted for potential confounders. We also modeled non-linear associations using restricted cubic splines. In secondary analyses, we restricted to acute lymphoblastic leukemia (ALL) cases and stratifed by age (<5 and ≥5 years). Compared with children who lived≥1000 m from a petrol station, the RR was 2.2 (95% CI 0.5-9.4) for children living<50 m from nearest petrol station. Associations were stronger for the ALL subtype (RR=2.9, 95% CI 0.6-13.4) and among older children (age≥5 years: RR=4.4, 95% CI 0.6-34.1; age<5 years: RR=1.6, 95% CI 0.1-19.4). Risk of leukemia was also greater (RR=1.6, 95% CI 0.7-3.3) among the most exposed participants when assigning exposure categories based on petrol stations located within 250 m of the child's residence and total amount of gasoline delivered by the stations. Overall, residence within close proximity to a petrol station, especially one with more intense refueling activity, was associated with an increased risk of childhood leukemia, though associations were imprecise.