Glacial surges in Svalbard are protracted and characterized by individual dynamic evolution, in contrast to many other areas, which calls for a subdivision of the classic two‐phased surge cycle. A dominating part of the ice masses seem to have a surge potential and this represents a considerable challenge for palaeoclimatic studies. Glaciological and geological models therefore need to be coupled. The issue is discussed with Fridtjovbreen glacier as an example. This ice mass is one of few glaciers studied throughout a surge cycle. It was active for 12 years (1991–2002) and represents the most protracted surge documented. The maximum advance rate was 4.2 m day−1, its maximum extent was reached after seven years, its run‐out distance was 4 km, and the relocated ice filled 5 km2 of the fjord. Intense subglacial thrusting occurred during various stages, including part of the ice‐front retreat, as shown by sub‐bottom profiling data from 2002. A six‐stage model is presented and processes are discussed with emphasis on the ice‐front retreat with transition to the quiescent phase. Although the surge mechanism itself is unrelated to climate, climatic conditions obviously play a major role in the course of a surge. During the surge, the ice mass made a dramatic impression in the landscape, but 10 years after the maximum extent, there is little onshore evidence of the event.