Exploiting admission thresholds to the Bologna day care system, we show using a regression discontinuity (RD) design that one additional day care month at age 0–2 reduces intelligence quotient by 0.5% (4.7% of a standard deviation) at age 8–14 in a relatively affluent population. The magnitude of this negative effect increases with family income. Similar negative impacts are found for personality traits. These findings are consistent with the hypothesis from psychology that children in day care experience fewer one-to-one interactions with adults, with negative effects in families where such interactions are of higher quality. We embed this hypothesis in a model that lends structure to our RD design.