The development of empirically-based rainfall thresholds for use in early warning and forecasting of landslides is a well established practice, with numerous studies from various regions around the world producing rainfall thresholds for a variety of landslide types. However, the identification of rainfall events which result in landslides is complex, particularly in areas with poor rainfall observations and regions affected by seasonal, monsoon-induced rainfall and highly variable spatial and temporal rainfall patterns. This study reviews the issues of rainfall event identification in these complex meteorological regions, using satellite-based precipitation estimates from the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Analysis Product (TMPA) and a newly collated regional landslide inventory for medium to large landslides in Papua New Guinea (PNG). The issues and challenges of identifying critical rainfall for historical landslides are assessed and a practical solution, for this data sparse region, is proposed.