Many agencies and researchers use data from harvested animals to study temporal trends in phenotype. For large mammals, complete harvest records are typically only available for the past few decades, but records of the largest trophies have been collected for over a century. To examine whether record books and data from male bighorn sheep (Ovis canadensis) harvested under a minimum-curl regulation could detect temporal trends in horn length, we simulated populations of trophy-harvested male bighorn sheep where horn length was modeled to increase, remain stable, and decrease over time. All populations experienced a simulated harvest based on a minimum horn length, but only horns in the longest 5% of the initial distribution were entered in a fictional record book. We then assessed whether monitoring of harvested and “record” males detected temporal trends. Data from selective harvest underestimated declines and initially underestimated increases, but qualitatively detected both trends. Record-book entries, however, severely underestimated increases and did not detect declines, suggesting that they should not be used to monitor population trends. When these biases are taken into account, complete trophy harvest records can provide useful biological information.