AbstractAt the River Bush salmon station in Northern Ireland, a genetically marked strain of Atlantic salmon,Salmo salarL., was established, with a low frequency (0.194) of a glucose‐6‐phosphate isomerase genotype (GPI‐3* 100/93). As part of a salmon enhancement programme, 43500 of these fish were stocked as swim‐up fry into a tributary of the nearby Margy River in Spring 1990, following a baseline genetic survey which indicated an absence of the GPI–3*93 allele in the wild population. This survey also indicated a significant frequency difference of a malate dehydrogenase allele (MDH‐B1,2*85) present both in the stocked and wild fish. A post‐stocking electrofishing survey in summer 1990 indicated higher summerling densities in stocked sections compared with unstocked (control) sections, with an estimated survival of stocked fry to summerling ranging from 24% to 29%. Genetic analysis based on GPI‐3*100/93 and MDH‐B1,2*100/85 genotypes provided similar estimates of the overall contribution of stocked fish at 59.8% to 61.9%, respectively, although there was a disparity between markers in estimated contributions to stocked and control sections. A further genetic survey in 1991 estimated that 32.2% (MDH) to 69.1% (GPI) of the 1 + parr population consisted of stocked fish, the MDH‐based estimate having declined greatly between summerling and 1 + parr stages. Possible reasons for the disparity of marker results are outlined, and the implications for use of low‐frequency gene