We read with interest the recent paper by Sariaslan et al.1 investigating the associations between childhood family income and three outcomes observed later in life: psychiatric disorders, substance misuse and violent crime arrests. The paper contains a careful analysis of a large and relevant dataset of more than 650 000 individuals born in Finland in 1986–96, who were followed from age 15, when income was assessed, until 2017 or 2018. The authors use two types of analysis in the paper: one in which differences in the outcomes are accounted for by variability in income between families (henceforth between-family analysis) and one, the so-called sibling comparison, in which outcome differences are accounted for only by income variability within families (henceforth within-family analysis). In the between-family analysis, the authors find consistent associations between a number of related estimates of childhood family income and the three outcomes (e.g. Figure 2 in the paper). However, and this is the main finding of the paper, these associations between income and the three outcomes completely disappear in the within-family analysis. From this, the authors conclude that ‘[a]ssociations between childhood family income and subsequent risks … were not consistent with a causal interpretation’.