Management programs for waterfowl populations include rationale for, and establishment of, hunting regulations. These programs rest partially on the results of scientific studies on the effect of harvest rates on annual survival rates. The evidence of this relationship has changed markedly since the mid-1970's, and it is not widely believed that a largely compensatory relationship exists between hunting mortality and other forms of mortality for the mallard (Anas platyrhynchos). This paper employs a general probabilistic model formulated to include a parameter (b) representing a continuum between complete compensation (b=0) and total additivity (b=1). Maximum likelihood estimates of this parameter were computer for 47 data sets of adult mallards banded throughout North American before hunting commenced. We found additional evidence of a highly compensatory mortality process for adult male mallards, while the evidence for adults female mallards remains inconclusive. Effective harvest, land acquisition, and land management programs depend upon additional information on the chronology and mechanisms underlying a compensatory mortality process.