To analyze the long-term variation in the spatial-temporal patterns of Antarctic krill (Euphausia superba) distribution, based on krill density data collected from 1926 to 2016, we analyzed the spatial and temporal variation of krill using the Getis-Ord Gi* statistic and the Triangulated Irregular Network (TIN) model. We compared the krill abundance, the percentage of krill abundance, and the size of hotspot/coldspot area. The results showed that there were one core hot-spot area, one sub-hotspot area and one edge hotspot area from 1926 to 1935 and from 1936 to 1945; one core hotspot area, one sub-hotspot area and one edge hotspot/coldspot area from 1976 to 1985; two core hotspot areas, two sub-hotspot areas, two edge hotspot areas, one sub-coldspot area and one coldspot area from 1986 to 1995; two core hotspot areas, two sub-hotspot/sub-coldspot areas and two edge hotspot/coldspot areas from 1996 to 2005; and from 2006 to 2016, there were two sub-hotspot areas, two edge hotspot areas, one sub-coldspot area and one edge coldspot area, respectively. Those hotspots/coldspots mostly occurred in the South Sandwich Islands, the Antarctic Peninsula, and the west of the Prydz Bay. From 1926 to 2016, krill abundance, the percentage of krill abundance in the hotspot area and the size of hotspot area had generally reduced, but the percentage of krill abundance in coldspot area and the size of coldspot area were increasing, while krill abundance in coldspot area decreased. It is a feasible approach to build an irregular triangle network to measure the size of hotspot/coldspot area and utilize hotspot analysis with 95% confidence to analyze the spatial-temporal pattern of krill distribution.