A gradual rise in sea level prior to a storm crossing the land is known as forerunner surge. While primary surge occurring around landfall due to wind set-up has been well understood, the nature of forerunner surge remains unclear. The aim of this research is to investigate generative mechanism, magnitude and phase speed of forerunner surge on Australian North West shelf based on historically large storms and tide gauge data. Five storms including Booby 1995, Vance 1999, Clare 2006, Glenda 2006 and Nicholas 2008 were selected in the analysis. These storms both initially were formed in the north, then travelled parallel to the coast, and finally made landfall spanning from King Bay to Exmouth with at an incident angle around 42⁰. Results indicated that the Ekman setup, a balance between Coriolis force and cross-shore pressure gradient, was the main driver of large forerunner surge in these events. The 40H low-pass sea levels showed that the magnitude of forerunner surge increases gradually from Broome to Onslow. The significant amplification of wave height from Cape Lambert to Onslow may involve simultaneous factors such as stormy parameters (storm speed, storm size, central air pressure), regional characters (shelf steepness), resonant conditions (phase speed, incident angle) and tide-surge interaction. After generating by Ekman set-up, forerunner surge migrates poleward as a forced shelf wave. Time lag analysis indicated that phase speed of the surge correlated closely with shelf width from Broome to Onslow. Particularly, a gradual decrease in speed propagation from 12 to 2 m/s corresponded to a narrowed shelf width between 195 and 80 km. When a storm moves into the shelf at angle around 42⁰, and translates slowly (about 4 ms^-1), a forerunner surge is likely to occur. This study is a fundamental step to improve better surge modelling on Australian North West shelf.