Population demography models are valuable tools in detecting population trends and identifying crucial components in the life history of a species. They are also useful in determining critical research needs to estimate needed parameters. A matrix-based p opulation model was expanded for Harlequin Ducks (Histrionicus histrionicus) originally developed by Goudie et al. (1994, Trans. North Am. Wildl. Nat. Resour. Conf. 59:27- 49.). A population growth rate of 16% (lambda=1.16) and a stable age distribution of 32.5% juveniles per year was calculated using parameters extracted from the literature. Both of these two values are much higher than seen in the Pacific Northwest populations (lambda ~= 1.00 and the proportion of juveniles ~= 10-15%). The population growth rate was most sensitive to survival of older age classes, however adult survival probably varies little on an annual basis. Juvenile survival (post-fledging to the following spring) and adult breeding propensity ha d a lesser effect on the population growth rate than adult survival, however, these two parameters are both, hard to estimate accurately, and, vary considerably between years. Lowering adult survival rates did not lower the proportion of juveniles in the projected population, unlike lowering juvenile survival and breeding propensity. Thus, the estimates of breeding propensity (0.6) and/or juvenile survival (0.5) must be higher than the actual field value. I decomposed the projection matrix into two matri ces, a breeding ground matrix and a non-breeding matrix to examine the effects of these periods separately. A substantial proportion of the sensitivity in the population growth rate occurs during the non-breeding period. Simulations using two different b reeding sites varying in quality revealed that movements of adults between sites had a greater impact on the population growth rate than juvenile movements. Better estimates for adult survival (breeding and non-breeding), juvenile survival and breeding pr opensity are currently being gathered and will be incorporated into future modeling efforts.