While recent studies mostly find that attending child care earlier improves the skills of children from low socio-economic and non-native backgrounds in the short-run, it remains unclear whether such positive effects persist. We identify the short- and medium-run effects of early child care attendance in Germany using a fuzzy discontinuity in child care starting age between December and January. This discontinuity arises as children typically start formal child care in the summer of the calendar year in which they turn three. Combining rich survey and administrative data, we follow one cohort from age five to 15 and examine standardised cognitive test scores, non-cognitive skill measures, and school track choice. We find no evidence that starting child care earlier affects children's outcomes in the short- or medium-run. Our precise estimates rule out large effects for children whose parents have a strong preference for sending them to early child care.