In this paper, the paired-gear and covered cod-end methods for estimating the selectivity of trawl cod-ends are compared. A modified version of the cod-end selectivity simulator PRESEMO is used to simulate the data that would be collected from a paired-gear experiment where the test cod-end also had a small mesh cover. Thus, estimates of the selectivity parameters of the test cod-end can be made using both the paired-gear method and the covered cod-end method. These estimates are compared and, as it is assumed that the covered cod-end method is objective, we conclude that the paired-gear method is biased. We demonstrate that extreme parameter estimates as well as discrepancies between the paired-gear and covered cod-end experiments do not necessarily reflect physical or biological mechanisms. We believe that this phenomenon may help explain cases in the literature where the covered cod-end and paired-gear methods produce different estimates of cod-end selectivity.