The current CCAMLR assessment method for mackerel icefish (Champsocephalus gunnari) in Subarea 48.3 employs the CCAMLR mixture analysis program (CMIX) and generalised yield model (GYM) packages, which derives population numbers-at-age and projects these numbers forward under the given harvest control rule to set a two-year total allowable catch (TAC). One issue is the accurate identification of age cohorts from the survey data. In this paper a length-based approach is described that removes the problem of cohort determination. A bootstrap technique (consistent with the current CCAMLR approach) is used to estimate the length distribution of the population from the survey data which, in conjunction with the bootstrapped survey biomass data, yields an estimate of the population numbers-at-length. A comparison of this method and the existing age-based CMIX/GYM methodology showed very good agreement in calculated yields for 2006, 2007 and 2009, and some differences for 2008 which were explained by an unusual length distribution from which it was difficult to reliably estimate age composition in CMIX. Using the 2009 survey data and the new methodology, a TAC was calculated according to the CCAMLR decision rules of 1 577 and 933 tonnes for the 2009/10 and 2010/11 fishing seasons respectively. The respective estimates using the existing CMIX/GYM methodology were 1 590 and 968 tonnes respectively.