On the basis of the bottom trawl surveys carried out by Russia and Great Britain during 2000, 2002, 2004 in the South Georgia area, it is demonstrated that the method by Aitchison and bootstrap method result in different estimates of mean fish density in the length series. Verification of the hypothesis of non-zero observations lognormal distribution for a large number of length groups casts doubt on the correctness of the first method application and therefore also of the method by de la Mare commonly used by CCAMLR in icefish age composition of abundance indices determination. To solve this problem it may be proposed to apply the modern version of the method by MacDonald and Pitcher. The comparison of this method and the method by de la Mare for length series, estimated with Aitchison’s method, were carried out. The results appeared very similar. The method by MacDonald and Pitcher has been tested on the length series obtained with the bootstrap method on the basis of the above mentioned surveys data. To estimate the standard error of the mixture distribution parameters, the “jackknifing-after-boostrap” method was proposed.