Objective: To evaluate the relationship between neonatal oxygen supplementation (O2) and childhood cancer in the Collaborative Perinatal Project (CPP). Study design: The CPP consisted of 54,795 children born between 1959 and 1966 and followed to age 8 years. We used Cox proportional hazards modeling to examine the association between history of neonatal O 2 and cancer (n = 48). Results: The hazard ratio (HR) for any O 2 was 1.77 (95% confidence interval [CI] = 0.94 to 3.35). The HR for continuous duration of O2 was near 1 and not significant. However, the HRs were 0.69 (95% CI = 0.17 to 2.88) and 2.87 (95% CI = 1.46 to 5.66) when comparing 0 to 2 and 3 or more minutes of O2, respectively, to no O2. The latter association was weaker (HR = 2.00; 95% CI = 0.88 to 4.54) and not significant (P = .10) when analysis was restricted to cancers diagnosed after age 1 year (n = 41). Conclusions: These findings are consistent with an association between O2 for 3 minutes or longer and cancer in childhood, and should serve as a basis for further study.