The ability to accurately recall locations and navigate our environment relies on multiple cognitive mechanisms. The behavioural and neural correlates of spatial navigation have been repeatedly examined using different types of mazes and tasks with animals. Accurate performances of many of these tasks have proven to depend on specific circuits and brain structures and some have become the standard test of memory in many disease models. With the introduction of virtual reality (VR) to neuroscience research, VR tasks have become a popular method of examining human spatial memory and navigation. However, the types of VR tasks used to examine navigation across laboratories appears to greatly differ, from open arena mazes and virtual towns to driving simulators. Here, we examined over 200 VR navigation papers, and found that the most popular task used is the virtual analogue of the Morris water maze (VWM). Although we highlight the many advantages of using the VWM task, there are also some major difficulties related to the widespread use of this behavioural method. Despite the task's popularity, we demonstrate an inconsistency of use - particularly with respect to the environmental setup and procedures. Using different versions of the virtual water maze makes replication of findings and comparison of results across researchers very difficult. We suggest the need for protocol and design standardisation, alongside other difficulties that need to be addressed, if the virtual water maze is to become the 'gold standard' for human spatial research similar to its animal counterpart.