Aim To investigate whether socio-economic status (SES) in childhood and school failure at 15 years of age predict illicit drug abuse in youth and young adulthood. Design setting and participantsRegister study in a Swedish national cohort born 1973-88 (n=1405763), followed from age 16 to 20-35 years. Cox regression analyses were used to calculate hazard ratios (HR) for any indication of drug abuse. Measurements Our outcomes were hospital admissions, death and criminality associated with illicit drug abuse. Data on socio-demographics, school grades and parental psychosocial problems were collected from censuses (1985 and 1990) and national registers. School failure was defined as having mean school grades from the final year in primary school lower than -1standard deviation and/or no grades in core subjects. Findings School failure was a strong predictor of illicit drug abuse with an HR of 5.87 (95% CI: 5.76-5.99) after adjustment for age and sex. Childhood SES was associated with illicit drug abuse later in life in a stepwise manner. The lowest stratum had a HR of 2.28 (95% CI: 2.20-2.37) compared with the highest stratum as the reference, when adjusted for other socio-demographic variables. In the fully adjusted model, the effect of SES was greatly attenuated to an HR of 1.23 (95% CI: 1.19-1.28) in the lowest SES category, while the effect of school failure remained high with an HR of 4.22 (95% CI: 4.13-4.31). Conclusions School failure and childhood socio-economic status predict illicit drug abuse independently in youth and young adults in Sweden.