A dataset for the 2009 assessment of Antarctic toothfish (Dissostichus mawsoni) in the Ross Sea was selected on the basis of data quality metrics for individual trips. Initial informative datasets were selected comprising trips with high (above median) rates of recovery of previously released tags, and/or where tags released on the trip were subsequently recaptured at a high rate. These trips were used to define a range for various data quality metrics considered to be informative with respect to tagging data. Other trips with data quality metric values within these ranges were added to the dataset. A retrospective analysis of this data selection method was undertaken to evaluate whether trip selection was stable over time. The 2009 dataset was reduced by omitting data from the two most recent years, and the data selection method applied to these 2007 and 2008 datasets. Results from this three year period indicate that additional, historical trips may be added to the data set as new tagging data become available. However, once selected, trips are also selected in subsequent years. This analysis indicates that the tagging data selection methodology developed for the 2009 assessment yields generally stable selections over time. The retrospective analysis period is, however, short, and stability of trip selection in future years as further data are added will also be of interest.