Early retirement options are usually targeted at employees at risk of not reaching their regular retirement age inemployment. An important at-risk group comprises older employees who have worked in demanding jobs formany years. This group may be particularly negatively affected by the abolition of early retirement options. Tomeasure differences in labor market reactions of employees in low- and high-demand jobs, we exploit the quasinaturalexperiment of a cohort-specific pension reform that increased the early retirement age for women from60 to 63 years. Based on a large administrative dataset, we use a regression-discontinuity approach to estimatethe labor market reactions. Surprisingly, we find the same relative employment increase of about 25% for treatedwomen who were exposed to low and to high job demand. For older women in demanding jobs, we also do notfind substitution effects into unemployment, partial retirement, disability pension, or inactivity. Eligibility forthe abolished early retirement option required high labor market attachment. Thus, we argue that this eligibilityrule induced a positive selection of healthy workers into early retirement. We propose alternative policies thatprotect workers exposed to high job demand better against the negative consequences of being unable to reachtheir statutory retirement age in employment.