Background One fifth of all deaths among children in Europe are accounted for by cancer. If this is to be reduced there is a need for studies on not only biology and treatment approaches but also on how social factors influence cure rates. We investigated how various socioeconomic characteristics were associated with survival after childhood cancer. Material and methods In a nationwide cohort of 3797 children diagnosed with cancer [hematological cancer, central nervous system (CNS) tumors, non-CNS solid tumors] before age 20 between 1990 and 2009 we identified parents and siblings and obtained individual level parental socioeconomic variables and vital status through 2012 by linkage to population-based registries. Hazard ratios (HR) and 95% confidence intervals (CI) for dying were estimated using multivariate Cox proportional hazard models. Results For all children with cancer combined, survival was slightly but not statistically significantly better the higher the education of the mother or the father, and with maternal income. Significantly better survival was observed when parents were living together compared to living alone and worse survival when the child had siblings compared to none. Young (<20) or older (≥40) maternal age showed non-significant associations, but based on small numbers. For hematological cancers, no significant associations were observed. For CNS tumors, better survival was seen with parents living together (HR 0.70, CI 0.51–0.97). For non-CNS solid tumors, survival was better with high education of the mother (HR 0.66, CI 0.44–0.99) compared to basic and worse for children with one (HR 1.45, CI 1.11–1.89) or two or more siblings (HR 1.29, CI 0.93–1.79) (p for trend 0.02) compared to none. Conclusion The impact of socioeconomic characteristics on childhood cancer survival, despite equal access to protocolled and free-of-charge treatment, warrants further and more direct studies of underlying mechanisms in order to target these as a means to improve survival rates.