Fish are increasingly popular subjects in behavioural and neurobiological research. It is therefore important that they are housed and handled appropriately to ensure good welfare and reliable scientific findings, and that species-appropriate behavioural tests (e.g. of cognitive/affective states) are developed. Routine handling of captive animals may cause physiological stress responses that lead to anxiety-like states (e.g. increased perception of danger). In fish, these may be particularly pronounced when handling during tank-to-tank transfer involves removal from water into air. Here we develop and use a new combined scototaxis (preference for dark over light areas) and novel-tank-diving test, alongside conventional open-field and novel-object tests, to measure the effects of transferring three-spined sticklebacks (Gasterosteus aculeatus) between tanks using a box or net (in and out of water respectively). Preference tests for dark over light areas confirmed the presence of scototaxis in this species. Open-field and novel-object tests failed to detect any significant differences between net and box-handled fish. However, the combined diving and scototaxis detected consistent differences between the treatments. Net-handled fish spent less time on the dark side of the tank, less time in the bottom third, and kept a greater distance from the ‘safe’ bottom dark area than box-handled fish. Possible explanations for this reduction in anxiety-like behaviour in net-handled fish are discussed. The combined diving and scototaxis test may be a sensitive and taxon-appropriate method for measuring anxiety-like states in fish.