Objectives Socio-economic position (SEP) and intelligence (IQ) are both predictors of later health, however, the relationships between them have been much debated. Our objective was to examine the separate effects of IQ and SEP on health, as measured by sickness absence (SA).

Methods All live-born males in Norway 1967–1971 (N = 170 678) were followed up in several national registers. Our study included subjects who were healthy at age 18–19 years, and at risk of SA at start of follow-up (N = 99 738). Health outcome was the 4-year risk (2000–2003) of at least one SA episode. IQ test scores were recorded at military conscription (age 18–19 years). Education level at age 28 and income in 2000 served as indicators of adult SEP, whereas parental education level and income during childhood were indicators of childhood SEP. Risk ratios (RRs) adjusted for birth year were estimated using generalised linear models. Direct and total effects were calculated according to our model.

Results A total of 23 506 subjects (24%) had SA episodes during follow-up. There were strong gradients according to participants’ IQ, educational attainment, and income. RRs between the lowest and highest of 5 categories were 5.99, 9.05, and 3.36, respectively. The effects were somewhat weakened when adjusted for childhood SEP, and were further reduced when IQ, education and income were all included in the analysis, to calculate the direct effects (RRs 1.84, 4.79, and 1.65, respectively). According to our model, the total effect of IQ (adjusted for childhood SEP), was RR 4.15.

Conclusions These results suggest that education had a stronger direct effect on health than IQ and income. However, the total effect of IQ was of similar strength and was in part mediated through adult SEP.