Landslides are one of the most widespread natural hazards in the world, responsible for thousands of deaths and billions of dollars in property damage every year. However, predicting landslide potential at a global scale using real-time monitoring is very difficult and expensive in terms of time and money. This is especially true in developing countries where expensive ground observation networks are prohibitive and in mountainous areas where access is minimize. This paper studies the relationship between rainfall and landslides event in Yogyakarta and Central of Java area. Landslides events and the triggering rainfall are collected from Kalibawang (Yogyakarta), Kebumen, Karanganyar, Banjarnegara (Central of Java). The landslide type at those locations is identified as shallow, deep-seated, and debris flow. Loss and damages of settlement and people was reported during the landslide occurrence. Based on the data collection from those landslide occurrences, an empirical rainfall threshold was determined based on intensity-duration (ID) and cumulative-intensity (EI) relationship. The threshold curve was obtained from the rainfall data using an objective statistical technique. The proposed ID and EI curves are also compared with the global existing rainfall threshold. The ID curve for the studied region is higher than the global rainfall threshold that proposed by several investigators. However, new regional rainfall thresholds for Yogyakarta and Central of Java can be used in a regional operational landslide warning system based on local or regional precipitation measurements where local and regional thresholds are not available.