Using rich panel data recently available from Spanish Social Security records, we find that a negative motherhood earnings differential of 2.3 log points remains even after controlling for both individual- and firm-level unobserved heterogeneity. The analysis of the mothers and childless women's earnings trajectories over time reveals that "mothers to be" experience important earnings increases (of up to 6 log points) several years prior to giving birth to their first child. However, this earnings' advantage gets seriously hit right after birth, and it is not until nine years later that mothers' earnings return to their pre-birth (relative) levels. The study finds that heterogeneity matters as most of the motherhood penalty and earnings' dip is driven by mothers working in the primary labor market (with permanent contracts). For these women, much of the earnings losses occur because mothers change employers to work part-time, or (if they stay with their former employer) they take leave of absence. An instrumental variable approach is used to address concerns of selection into type of contract. We exploit variation in the amount, timing and profiling of subsidies offered to firms when hiring permanent workers, a policy that started to be implemented in Spain in 1997.