We have developed a model to test several underlying assumptions of bacterial source tracking (BST) when the BST method is based on detection of discrete genetic markers from source-specific bacteria. The model consists of an environment and discrete-time input signals that represent sources of contamination partitioned into marker-bearing and nonmarker-bearing units "shed" into the environment. Simulations run for different types of environmental contamination patterns indicate that if hosts shed different percentages of BST markers, the environment cannot be accurately characterized unless a correction method is used. The correction method, which requires the solution to a linear system, reduces the mean error in estimating the proportions of host contamination to below 3%. The effectiveness of the method depends on accurate knowledge of the occurrence and prevalence of markers in the various hosts; this may be a challenging task, especially if these values vary across populations in space and time. In addition, the correction method does not compensate for environments with low-density or unmixed contamination. In conclusion, our simulations highlight several fundamental challenges that may prevent absolute quantification of fecal input using discrete marker BST.