Line transect analysis was applied to the 1987-1990 North Pacific marine mammal sightings survey data, with a regression adjustment for size-biased sampling of schools. Abundance of Pacific white-sided dolphins is estimated as 931,000 animals (cv = 0.900), with 95 % confidence interval (206,000 – 4,216,000), and abundance of northern right whale dolphins is estimated as 68,000 animals (cv = 0.709), with 95 % confidence interval (20,000 – 239,000). Similar methods applied to Dall’s porpoise yield an estimate of 1,186,000 animals (cv=0.092), with 95 % confidence interval (991,000 -1,420,000). Data on northern fur seals proved easier to model, and abundance is estimated as 190,000 animals (cv= 0.149), with 95 % confidence interval (142,000 -254,000). If animals are attracted towards the survey before detection, then positive bias is potentially large; evidence for such behaviour is strong, especially for Pacific white-sided dolphins. If some schools on the trackline are undetected during periods when the observer is on duty, then estimates are also affected by a negative source of bias. From the available data, it is not possible to quantify these sources of bias.