Armstrong (1995) presented a final report of scientific observations conducted aboard the Japanese krill fishing vessel Chiyo Maru No. 2. Unfortunately, Armstrong's fish by-catch data were not presented in a standardised manner. This paper presents a brief re-analysis of the by-catch data collected on board the Chiyo Maru No. 2; means and variances are calculated from the delta distribution. In numbers, an average of about 14 fish were caught per tome of krill, and an average of about 25 fish were caught per hour of towing. In weight, an average of about 1.2 kg of fish were caught per tonne of krill, and an average of about 2.2 kg of fish were caught per hour of towing. Confidence intervals for mean by-catch estimates are wide suggesting that, in the future, more hauls should be sampled for by-catch. The proportion of hauls containing fish was much higher for the Chiyo Maru No. 2 than it was for other Japanese and Ukrainian krill trawlers fishing in the Atlantic sector of the Southern Ocean. This difference may be related to differences in the amount of krill that was subsampled for making observations on fish by-catch. The by-catch data from the Chiyo Maru No. 2 did not provide a clear picture of the reiationship between haul-specific fish by-catch and the catch rate of krill, and this observation is different from those made in most other studies. It would be worthwhile to estimate an appropriate sample size (in numbers of hauls) and subsample size (as a fraction of haul-specific krill catches) for observers collecting data on fish by-catch.