Because unionid mussels have a parasitic larval stage, they are able to disperse upstream and downstream as larvae while attached to their host fish and with flow as juveniles after excystment from the host. Understanding unionid population ecology requires knowledge of the processes that affect juvenile dispersal prior to establishment. We examined presettlement (transport and dispersion with flow) and early postsettlement (bed shear stress) hydraulic processes as negative censoring mechanisms. Our approach was to model dispersal using particle tracking through a 3-dimensional flow field output from hydrodynamic models of a reach of the Upper Mississippi River. We tested the potential effects of bed shear stress (??b) at 5 flow rates on juvenile mussel dispersal and quantified the magnitude of these effects as a function of flow rate. We explored the reach-scale relationships of Froude number (Fr), water depth (H), local bed slope (S), and unit stream power (QS) with the likelihood of juvenile settling (??). We ran multiple dispersal simulations at each flow rate to estimate ??, the parameter of a Poisson distribution, from the number of juveniles settling in each grid cell, and calculated dispersal distances. Virtual juveniles that settled in areas of the river where b > critical shear stress (c) were resuspended in the flow and transported further downstream, so we ran simulations at 3 different conditions for ??c (??c = ??? no resuspension, 0.1, and 0.05 N/m2). Differences in virtual juvenile dispersal distance were significantly dependent upon c and flow rate, and effects of b on settling distribution were dependent upon c. Most simulations resulted in positive correlations between ?? and ??b, results suggesting that during early postsettlement, ??b might be the primary determinant of juvenile settling distribution. Negative correlations between ?? and ??b occurred in some simulations, a result suggesting that physical or biological presettlement processes might determine juvenile settling distributions. Field data are needed to test these hypotheses. Results support the idea that flow patterns and b can act as negative censoring mechanisms controlling settling distributions. Furthermore, a river reach probably has a quantifiable threshold range of flow rates. Above the upper threshold, ??b probably is the primary determinant of juvenile settling distribution. Relationships of ?? with H, Fr, S, and QS were relatively weak. Important physical processes that affect dispersal probably are not captured by approximations based on large-scale hydraulic parameters, such as Fr and H. ?? 2010 The North American Benthological Society.