Enrichment of the environment for captive animals is aimed at producing beneficial effects on the behaviour and physiology of relevant species, and is commonly used to reduce harmful social behaviours and stereotypies. However, little work has been undertaken to develop enrichment strategies for fish, in particular those used in regulatory toxicology where strict criteria regarding holding conditions and experimental design can make implementing such techniques problematic. Here, we studied the effect of vertical rod structures, designed to increase environmental complexity and provide refuge, on several commonly cited anxiety-related behaviours and whole-body levels of the stress hormone cortisol in juvenile zebrafish measured over a 1-week period. Activity levels and shoaling density showed no response to tank structures and fish did not spend a significantly greater or lesser amount of time in areas of tanks containing glass rods. Aggression remained high during days 1-5 in tanks containing glass structures before falling to a lower level by day 7. In control tanks, this lower level was reached 2 days earlier, by day 5, suggesting that the glass structures may have slowed the rate of establishment of dominant/subordinate relationships. Overall, whole-body cortisol levels of fish were comparable to those reported in unstressed zebrafish in other studies. Levels were significantly higher in both treatments after 24 h than on subsequent days, most likely due to the handling stress of the initial transfer to experimental tanks. However, cortisol levels did not vary significantly between control and structured tanks at any point during the study. These results indicate that the addition of glass rod structures as hypothesised enrichment did not result in a measurable improvement in welfare.