Research has recently established the notion that humor in leadership contributes to the development of a positive professional relationship between leaders and followers. This relationship has been supposed to be the core mechanism via which humor in leadership unfolds its effects on work attitudes and behaviors. However, research has neglected the option that humor used by leaders might fail to amuse their followers. In this study, we investigate the role of failed humor for the relationship between leader and follower. More concretely, we develop a new scale for measuring failed humor in leadership and demonstrate its factorial and criterion-related validity. Using an automated item selection algorithm, we optimized the newly developed scale and derived a well-fitting six-item scale out of a pool of 12 items. In a study based on a sample of 385 employees, we were able to show that our newly developed scale is factorially valid. Moreover, we showed a negative correlation between failed humor and leader-member exchange. Furthermore, we showed incremental validity of failed humor in that failed humor predicted variance in leader-member exchange beyond well-established humor constructs such as affiliative and aggressive humor. Our study contributes to the development of the field of humor in leadership and opens up new options for further inquiry. Moreover, our study demonstrates the use of automated item selection algorithms in the applied field.