Many animals engage in elaborate behavioural sequences in order to build structures. The ‘end-products’ (e.g., nests of birds, cases of caddisflies) of these behaviours have occasionally been used either to infer details of building behaviour or to infer that structures similar in appearance are constructed by similar behavioural sequences. However, behavioural information can be extrapolated from end-products only if there is congruence between the end-product structure and the animal's movements. Case building behaviour was studied in caddisflies (Trichoptera) to examine the relationship between end-products and behaviour. We found examples of taxa that built similar structures using different behavioural patterns, and taxa that built different structures using similar behaviours, regardless of whether these taxa were closely or distantly related. These findings question the reliability of behavioural inferences based solely on details of their associated end-products and suggest that end-product structure should be removed from the definition of animal behaviour.