I analyzed data from the 1984 and 1986 Federal Waterfowl Hunter Questionnaire Survey (WHQS) to estimate the rate of return of name and address contact cards, to evaluate the efficiency of the Survey's stratification scheme, and to investigate potential sources of bias due to nonresponse at the contact card and questionnaire stages of the Survey. Median response at the contact card stage was 0.200 in 1984 and 0.208 in 1986, but was lower than 0.100 for many sample post offices. Large portions of the intended sample contributed little to the final estimates in the Survey. Differences in response characteristics between post office size strata were detected, but size strata were confounded with contact card return rates; differences among geographic zones within states were more pronounced. Large biases in harvest and hunter activity due to nonresponse were not found; however, consistent smaller magnitude biases were found. Bias in estimates of the proportion of active hunters was the most pronounced effect of nonresponse. All of the sources of bias detected would produce overestimates of harvest and activity. Redesigning the WHQS, including use of a complete list of waterfowl hunters and resampling nonrespondents, would be needed to reduce nonresponse bias.