We estimate the causal effects of childcare availability on the maternal employment rate using prefecture panel data constructed from the Japanese quinquennial census 1990-2010. We depart from previous papers on Japan by controlling for prefecture fixed effects, without which the estimates can be severely biased upward. Contrary to popular belief, childcare availability is uncorrelated with maternal employment when prefecture fixed effects are controlled. Evidence suggests that this is because households shift from using informal childcare provided by grandparents to the accredited childcare service, as more and more households do not live with grandparents. If this change of the household structure did not occur, the growth of childcare availability would have increased the maternal employment rate by two percentage points, which accounts for about 30% of the growth in the maternal employment rate from 1990 to 2010. (author abstract)