The collapse of the northern Atlantic cod (Gadus morhua) fishery off southern Labrador and to the northern Grand Bank of Newfoundland, once the largest cod fishery in the world, was a social and economic disaster for the region. An analysis of traditional catch-at-age data in conjunction with research surveys, which assumed that research survey estimation errors of abundance by age and year were independent, led assessment biologists to the conclusion that the collapse was caused by an increase in natural mortality in the first half of 1991. We constructed a statistical model to test this hypothesis. The results do not support the hypothesis. There is ambiguous evidence that natural mortality has increased since 1991; however, these results are found only in a model that has extraordinary patterns in the residuals. Our analysis suggests that even if natural mortality has been higher in recent years (as estimated using a model with correlated errors for research surveys), overfishing was sufficiently high to cause a collapse of this population. We also demonstrate that the usual assumption that estimation errors from research trawl surveys are independent is not valid, and can lead to invalid inference and unreasonable estimates of abundance.