Relocation of unionid mussels into refuges (e.g., hatchery ponds) has been suggested as a management tool to protect these animals from the threat of zebra mussel (Dreissena polymorpha) invasion. To evaluate the efficacy of relocation, we experimentally relocated 768 mussels, representing 5 species (Leptodea fragilis, Obliquaria reflexa, Fusconaia flava, Amblema plicata, and Quadrula quadrula) into an earthen pond at a National Fish Hatchery or back into the river. In both locations, mussels were placed into 1 of 4 treatments (mesh bags, corrals, and buried or suspended substrate-filled trays). Mussels were examined annually for survival, growth (shell length and wet mass), and physiological condition (glycogen concentration in foot and mantle and tissue condition index) for 36 mo in the pond or 40 mo in the river. We observed significant differences in mortality rates between locations (mortality was 4 times greater in the pond than in the river), among treatments (lowest mortality in the suspended trays), and among species (lower mortality in the amblemines than lampsilines). Overall survival in both locations averaged 80% the 1st year; survival in the pond decreased dramatically after that. Although length and weight varied between locations and over time, these changes were small, suggesting that their utility as short-term measures of well being in long-lived unionids is questionable. Mussels relocated to the pond were in poor physiological condition relative to those in the river, but the magnitude of these differences was small compared to the inherent variability in physiological condition of reference mussels. These data suggest that relocation of unionids into artificial ponds is a high-risk conservation strategy; alternatives such as introduction of infected host fish, identification of mussel beds at greatest risk from zebra mussels, and a critical, large-scale assessment of the factors contributing to their decline should be explored.