High-density pilgrimage destinations face a tension between accommodating growing visitor numbers and preserving the safety, spiritual quality and heritage value of sacred sites. This paper applies agent-based visitor flow modelling to two high-density pilgrimage destinations, the Grand Mosque precinct in Mecca during non-Hajj Umrah periods and the Fushimi Inari and Kiyomizu-dera pilgrimage corridors in Kyoto, in order to evaluate management measures for reducing congestion. The models represent individual visitors as agents with heterogeneous origins, itineraries, walking speeds, group sizes and responsiveness to information, moving through spatially explicit networks calibrated with pedestrian counts, mobile location data and observed dwell times collected in 2025. Four management measures were tested individually and in combination: time-slot reservation, real-time information nudges delivered through mobile applications and signage, physical route separation, and capacity capping at entrances. Performance was assessed using peak density, mean waiting time, throughput and a satisfaction index derived from survey-based relationships between crowding and visitor experience. Scenario analysis identifies time-slot reservation and information nudges as the most effective measures for reducing congestion without lowering visitor satisfaction, with the combined scenario reducing peak densities by 34 per cent in Mecca and 29 per cent in Kyoto while holding satisfaction within two points of the baseline. Capacity capping reduced density comparably but lowered satisfaction because of extended waiting, and route separation produced modest gains that depended on site geometry. The paper discusses the transferability of measures between destinations with different religious, regulatory and cultural characteristics and proposes a decision framework for sustainable pilgrimage tourism management.