We conducted a reward band study on mallards (Anas platyrhynchos) to estimate and test hypotheses about sources of variation in band reporting rate. We banded 25,398 mallards with standard and 100 reward bands (3 mallards banded with standard bands for every reward-banded mallard) during preseason (Jul-Sep), 1988. We used a series of multinomial models to model the resulting 2,776 band recoveries from 1988 to 1991. Estimates of reporting rate for males shot in 10 harvest areas ranged from 0.29 to 0.46 and averaged 0.38 ( cxa SE = 0.020). We found evidence (P lt 0.01) of geographic variation in reporting rates, but not of smooth latitudinal or longitudinal gradients. There was evidence (P = 0.07) of lower reporting rates for females than males, especially in prairie Canada and the Central Flyway. Except for young males in the northern Atlantic Flyway, estimated harvest rates were lower than historical estimates, as expected from recent restrictive hunting regulations. Patterns of geographic and agesex variation in harvest rates were similar to those obtained using historical band-recovery data.