Summary1. Gene flow from domesticated to wild populations is a major threat to wild salmonid fish. However, few studies have addressed how populations could be restored after admixture has occurred. We analysed the prospects for restoring the previously intensively stocked brown trout population of the Skjern River, Denmark, by identifying remaining nonadmixed individuals to be used for supportive breeding. 2. We analysed microsatellite DNA markers in historical (1940 -50s) and contemporary (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004) samples from the Skjern River system, from the strain of domesticated trout previously used for stocking, and from the neighbouring Storå River. We analysed admixture proportions to estimate the genetic contribution by domesticated trout. We identified non-admixed trout using assignment tests, and further analysed the possible sources of indigenous trout by estimating contemporary migration among populations. 3. Genetic differentiation between the historical Storå and Skjern river populations was low ( θ ST = 0·004), suggesting considerable gene flow in the past. The contemporary Skjern and Storå river populations and a supportive breeding brood stock were strongly admixed, but some non-admixed individuals nevertheless remained in the wild-caught samples. In addition, two resident populations in isolated tributaries were found to be indigenous. The indigenous anadromous individuals from the Skjern River were unlikely to have been recruited from either the isolated tributary populations or the neighbouring Storå River and were presumably derived from unidentified spawning sites in the river system. 4. All but one non-admixed anadromous Skjern River trout were females, which we ascribed to sampling bias. Moreover, all non-admixed fish were late-spawning (JanuaryFebruary) whereas the majority of all trout caught for the study were ripe by NovemberDecember. The difference in spawning time could be an important factor delaying complete admixture of domesticated and indigenous trout. 5. Synthesis and applications. This study demonstrates the feasibility of restoring populations that have been admixed with exogenous individuals, by identifying non-admixed individuals using genetic markers. However, the results also highlight the problem that numbers of identified non-admixed individuals may be small, necessitating identification of nearby, closely related populations that can be incorporated into breeding programmes.