When correlating brain size and structure with behavioural and environmental characteristics, a range of techniques can be utilised. This study used gobiid fishes to quantitatively compare brain volumes obtained via three different methods; these included the commonly used techniques of histology and approximating brain volume to an idealised ellipsoid, and the recently established technique of X-ray micro-computed tomography (micro-CT). It was found that all three methods differed significantly from one another in their volume estimates for most brain lobes. The ellipsoid method was prone to over- or under-estimation of lobe size, histology caused shrinkage in the telencephalon, and although micro-CT methods generated the most reliable results, they were also the most expensive. Despite these differences, all methods depicted quantitatively similar relationships among the four different species for each brain lobe. Thus, all methods support the same conclusions that fishes inhabiting rock pool and sandy habitats have different patterns of brain organisation. In particular, fishes from spatially complex rock pool habitats were found to have larger telencephalons, while those from simple homogenous sandy shores had a larger optic tectum. Where possible we recommend that micro-CT be used in brain volume analyses, as it allows for measurements without destruction of the brain and fast identification and quantification of individual brain lobes, and minimises many of the biases resulting from the histology and ellipsoid methods.