Various elements of the Generalized Linear Model (GLM) used to standardise toothfish CPUE in Subarea 48.3 are investigated. Problems with crossover between the essentially summer fishery of 1992-1993 prosecuted by Russian vessels and the winter fishery of 1995 onwards prosecuted by other nationalities, coupled with the existence of the experimental fishery in 1994, lead us to suggest that future analyses should be restricted to GLMs based on winter data alone. Unfortunately, this does result in there being no data for 1993, one of the most successful summer seasons. The relationship between CPUE and depth is also explored, and tins leads to some suggestions about the distribution of toothfish of different sizes on the shelf slope. A winter GLM is described which includes a depth factor, and is used in a discussion of the recent declines in CPUE seen in Subarea 48.3.