Mackerel icefish, Champsocephalus gunnari (Lönnberg), at South Georgia and Shag Rocks in the Southern Ocean have supported a fishery since the 1970s. This study has analysed the length-frequency distribution of C. gunnari from 10 bottom trawl surveys at South Georgia and 9 at Shag Rocks between 1987 and 2002. Most surveys were between December and February (summer), with one in September (Spring). Mean total lengths (TL) of age-classes were determined for each year by examining length frequency plots and tables, and by using CMIX. These methods were only applied where there was distinct modality in the length-frequency distributions. Lengths of age-classes derived by these two methods were not statistically different. Age-class 1 was found for all surveys at South Georgia but was absent for 3 years at Shag Rocks, potentially indicating greater recruitment variability at Shag Rocks. Age-class 4 was found for 8 surveys at South Georgia but for only 1 survey at Shag Rocks, indicating likely differences in mortality between localities. To compensate for variation in the time of surveys, the mean TL of age-classes were standardized to a common day of the year (16 January) based on the Bertalanffy growth curve. The CMIX estimated mean TL of age-classes 1, 2 and 3 were, respectively, 14.5, 23.8 and 30.2 cm at South Georgia, and 18.6, 26.8 and 33.6 cm at Shag Rocks. The mean TL of each age-class of C. gunnari at Shag Rocks was significantly larger than at South Georgia, although the annual growth increment at each locality was similar. This is further evidence that C. gunnari probably hatch earlier at Shag Rocks. A difference in hatching period between the two localities and differences in recruitment and mortality indicates that the mackerel icefish populations at South Georgia and Shag Rocks should be managed as two different stocks. At South Georgia, the mean TL of age-class 1 decreased significantly between 1987 and 1994, and this change was negatively correlated with summer sea surface temperatures during the previous year. Summer maximum sea surface temperatures at South Georgia have increased significantly between 1950 and 2000, and this shift in temperatures is likely to have changed the seasonal timing and level of primary production. The decreased size of C. gunnari may be the result of reduced nearshore food availability linked to climate variability.