The pros and cons of using anaesthesia when handling fish in connection with experiments are debated. A widely adopted practice is to wait thirty minutes after anaesthesia before behavioural observations are initiated, but information about immediate effects of a treatment is then lost. This is pertinent for responses to acute stressors, such as acid injection in the acetic acid pain test. However, omission of anaesthetics in order to obtain data on immediate responses will compromise the welfare of fish and contribute to experimental noise due to stress. We therefore tested the effect of tricaine methanesulfonate on the behaviour of zebrafish. We predicted that tricaine (MS 222) would decrease swimming velocity and that the control fish would show an increased level of anxiety- and stress-related behaviours compared to the tricaine group. Following acclimatization to the test tank, baseline behaviour was recorded before immersion in either tricaine (168 mg l(-1), treatment group, N = 8) or tank water (control group, N = 7). Latencies to lose equilibrium and to lose response to touch were registered. The fish was then returned to the test tank, and the latency to regain equilibrium was registered in anaesthetized fish. When equilibrium was regained, and at five, thirty and sixty minutes after the fish had been returned to the test tank, behaviour was recorded. The tricaine fish showed the following responses (mean ± sd): latency to lose equilibrium 22.6 s±3.9; latency to lose response to touch 101.9 s±26.8; latency to regain equilibrium 92.0 s±54.4. Contrary to our predictions, neither treatment caused a change in any of the behaviours registered. This indicates that tricaine has no effect on several commonly used behavioural parameters, and that it may be unnecessary to postpone behavioural observations to 30 min after anaesthesia.