We analyse efficiency of a labour market matching process at different levels of territorial aggregation and in different time horizons. We apply a stochastic matching frontier to random, job queuing and stock‐flow models. We use data for Poland, a country with highly regionally diversified unemployment. We find that search and matching mechanisms significantly differ with territorial aggregation. Higher efficiency in small regions results from lower frictions. Larger regions enjoy better structural matching if only labour market participants have time to search. Different policy measures should be designed to improve labour market matching efficiency at different regional levels.