A study was carried out in the River Kerry, north-western Scotland, in order to determine the sampling efficiencies for different size classes of endangered freshwater pearl mussels (Margaritifera margaritifera). An experienced worker searched 20 x 1 m² quadrats on the river bed in order to recover secretly hidden mussel samples (n = 2-74 mussels per quadrat). An overall recovery rate of 0.69 (= 269/390 hidden mussels) was achieved. A significant trend of increasing sampling efficiency with mussel size was observed. Our results suggest that the lack of juveniles observed in some, otherwise 'healthy', M. margaritifera populations may be explained by sampling bias.