Kirkwood and Agnew (2001, paper WG-FSA-01/48) proposed an ad hoc method for estimating selectivity curves for toothfish taken by longliners around South Georgia and Shag Rocks over the period 1995 – 1999. This method was applied by WG-FSA-01, with minor amendments, to estimate annual age-specific selectivities for the years 1995 – 2000 for input to the GYM assessment. In this paper, I propose a revised method for estimating the selectivities based on an assumption that the proportions of the total CPUE in an area for a particular length class that are taken in different depth zones are Beta-distributed. This removes some of the ad hoc nature of the former estimation method. The method is illustrated by application to data for 1995 – 2000.