Harvest data (e.g., number of animals harvested, trapper effort) are an important source of information for state wildlife agencies to manage harvested furbearers. These data provide evidence to support adapting harvest regulations when necessary. Setting appropriate harvest regulations for fishers (Martes pennanti) and American martens (Martes americana) is critical, as these species often exist at low densities, are sensitive to timber-management practices and trapper-harvest, and experience some level of interspecific predation and competition in sympatric populations. We estimated effects of management (e.g., number of fishers or martens harvested per trapper per season [harvest limit], season length) and extrinsic (e.g., weather, pelt prices) factors on regulated harvests of fishers and martens in the Upper Peninsula of Michigan during 1996–2007. We used generalized linear mixed models in an information-theoretic approach (quasi-likelihood adjusted Akaike Information Criterion [QAIC]) to discern which factors most strongly influenced fisher and marten harvests. For harvest of fishers, the 3 QAIC-best models included harvest limit, season length, and number of trappers, suggesting that regulatory changes within the ranges tested may be implemented to influence harvest. The QAIC-best model (harvest limit) contained 26% of the weight of evidence, and using an independent subset of data, showed no difference between model predictions and harvest data. In contrast, harvest of martens was not strongly influenced by any factors we tested. Possible reasons for a lack of measurable effects while modeling harvest of martens include a low harvest limit (i.e., 1 marten) or incidental harvest of martens by fisher or bobcat (Lynx rufus) trappers. Knowledge of influences on harvest will lead to informed decision-making when managers are setting harvest regulations, particularly for low-density furbearers.