In the rat, fear-potentiated startle (FPS) test animals are first trained to associate brief light presentations with a mild electric footshock and then tested for startle responses to acoustic stimuli, delivered either in darkness (i.e. baseline startle) or after the conditioning stimulus. Following light presentation the magnitude of the startle response is markedly increased, and the test is commonly used to distinguish anxiolytic drug effects (i.e. a reduction in FPS) from non-specific effects such as sedation/muscle relaxation. However, recent studies suggest that the environment in which the animal is trained may also contribute towards the acquisition of a conditioned fear response (i.e. contextual fear conditioning) and that this may elevate startle responses recorded in the dark. In the present study, therefore, we have compared the benzodiazepine/γ-aminobutyric acid-A receptor agonist chlordiazepoxide with the partial agonists FG 8205 and bretazenil, which are known to have a reduced propensity to produce sedation/myorelaxation, using two different FPS procedures: (i) conditioning and testing in stabilimeter chambers, and (ii) conditioning and testing in different environments. The results show that FPS can be demonstrated in both procedures and that treatment with chlordiazepoxide, FG 8205 or bretazenil dose-dependently attenuates the response. However, animals conditioned and tested in stabilimeter chambers also showed a significant increase in dark-startle amplitudes compared with non-shocked rats, suggesting that this response was elevated by contextual fear conditioning. Furthermore, despite clear differences in side-effect liabilities, FG 8205 and bretazenil significantly reduced dark-startle responses, suggesting that this measure is also sensitive to the anxiolytic effects of benzodiazepines. In contrast, when animals were conditioned and tested in different environments, dark-startle responses were not significantly different from those recorded in non-shocked rats and treatment with FG 8205 or bretazenil had no effect. Thus, conditioning and testing animals in different environments may provide a more effective means of distinguishing anxiolytic from non-specific drug effects in the rat FPS test.