Chinook Salmon Oncorhynchus tshawytscha sport fisheries in the Columbia River basin are commonly monitored using roving creel survey designs and require precise, unbiased catch estimates. The objective of this study was to examine the relative bias and precision of total catch estimates using various sampling designs to estimate angling effort under the assumption that mean catch rate was known. We obtained information on angling populations based on direct visual observations of portions of Chinook Salmon fisheries in three Idaho river systems over a 23-d period. Based on the angling population, Monte Carlo simulations were used to evaluate the properties of effort and catch estimates for each sampling design. All sampling designs evaluated were relatively unbiased. Systematic random sampling (SYS) resulted in the most precise estimates. The SYS and simple random sampling designs had mean square error (MSE) estimates that were generally half of those observed with cluster sampling designs. The SYS design was more efficient (i.e., higher accuracy per unit cost) than a two-cluster design. Increasing the number of clusters available for sampling within a day decreased the MSE of estimates of daily angling effort, but the MSE of total catch estimates was variable depending on the fishery. The results of our simulations provide guidelines on the relative influence of sample sizes and sampling designs on parameters of interest in short-duration Chinook Salmon fisheries.