This study investigates the impact of women's education on fertility. For identification, we use the 1968 compulsory education law change in Taiwan, which generated a regression discontinuity design (RDD) setting. We use the whole population of women from the 1980 and 2010 Population Censuses. Results of our RDD estimation using the exact date of birth suggest that the law change was effective in boosting women's education, but it did not have any impact on fertility. This is in stark contrast to most previous studies using only the birth year as the running variable or using it to construct instruments, which find that women's education depresses fertility. This study demonstrates that using a discrete running variable in RDD may generate a false discontinuity for an otherwise continuous regression function. (JEL J13, C21, I2)