Reservoir releases may be specified to flush interstitial fine sediment from gravel beds in the river downstream. Choice of an effective flow depends on trade-offs among discharge, flow duration, and pool dredging as they determine rates of bed mobilization, sand removal, and gravel loss. A basis for evaluating these trade-offs is developed with an approximate method appropriate to the sparse data typically available. Sand and gravel transport are represented with rating curves. Approximate methods are introduced for estimating effective gravel entrainment, subsurface sand supply, and pool sediment trapping. These are combined in a sand routing algorithm to evaluate flushing alternatives for the Trinity River, California. A sediment maintenance flow of moderate size, just sufficient to entrain the bed surface over the duration of the release, limits gravel loss and maximizes sand trapping by pools. Larger discharges produce more fines removal but at the cost of greater gravel loss and reduced selective transport of fines. Dredged pools increase sand removal efficiency by providing multiple exits from the channel and minimize gravel loss if dredged sediment is screened and gravel returned to the river. ¿ American Geophysical Union 1996