Evidence from neuropsychology suggests that the distinction between animate and inanimate kinds is fundamental to human cognition. Previous neuroimaging studies have reported that viewing animate objects activates ventrolateral visual brain regions, whereas inanimate objects activate ventromedial regions. However, these studies have typically compared only a small number of animate and inanimate kinds (e.g. animals and tools) and some evidence indicates that task demands determine whether these effects occur at all. In the current study we test whether a lateral-medial animacy bias is evident across a variety of stimuli, and across different tasks (matching two stimuli at a general, intermediate and exemplar level). Images of objects were presented sequentially in pairs, and match/mismatch judgements were made at different levels in different scans. The fMRI data showed ventrolateral activation for animate objects and ventromedial activation for inanimate objects. Additional analyses within these regions revealed no main effect of task, and no interactions between task and animacy. Furthermore, there were no subpopulations of voxels in any of the regions of interest that showed a significant task by animacy interaction. We conclude that ventral animate/inanimate category biases do not always depend on top-down task orientation. Furthermore, we consider whether the animate and inanimate activations reflect biases in the non-preferred responses of strongly category-selective regions such as the fusiform face area or the parahippocampal place area. (Contains 4 figures and 2 tables.)