As many courts have struggled with an increasing workload and a backlog of cases, it might come as no surprise that some have implemented a reduction in panel size to improve the efficiency and cost-effectiveness of decision making. Yet, we have only very limited empirical evidence as to if and how reducing panel size affects judicial decisions. This paper leverages a natural experiment at the Swiss Federal Administrative Court to gauge the impact of an effective panel size reduction on asylum appeal decisions. Drawing on fuzzy regression discontinuity design and machine learning approaches, I show that giving chair judges the opportunity to invoke a shorter procedure involving one less judge in the decision of cases ‘clearly with or without merit’ leads to a decrease in the grant rate. Though perhaps particularly consequential in the field of asylum law, this finding draws attention to the importance of decision-making procedures for the consistency of judicial decisions more broadly.