This study presents a new field experimental approach for measuring age discrimination in hiring. In addition to the classical approach in which candidates' ages are randomly assigned within pairs of fictitious resumes that are sent to real vacancies, we randomly assign activities undertaken by the older candidates during their additional life years between these pairs. When applying this design to the Belgium case, we find that age discrimination is fundamentally heterogeneous by older candidates' career pattern. Older age affects call-back only (negatively) in case older candidates were inactive or employed in an out-of-field job during their additional post-educational years.