Neuroimaging has revealed a left-lateralized network of brain areas implicated in understanding the conceptual and sensorimotor aspects of tool perception and tool use. Often this network of areas is identified by contrasting brain activity when participants view pictures of tools vs. pictures of non-tools (e.g., animals or buildings). It is unclear, however, what aspect of tools drive activity in the tool network as both tools and non-tools tend to differ in their low-level features. For instance, areas in the tool network may simply activate to elongated objects or to handheld objects over round or ungraspable objects irrespective of object category. To test whether tools indeed drive activity in tool-selective areas over non-tools, participants passively viewed real tools and non-tools matched on low-level features during functional magnetic resonance imaging (fMRI). To maximize the potential for action, participants saw real-tools as opposed to pictures of tools. The non-tools were created by chopping the business ends of tools into pieces and attaching the pieces to both ends of the original tool handles. In doing so, the tools and non-tools were matched for elongation and real-world size. Importantly, tools and non-tools were viewed directly without the use of mirrors and placed within the participants' reach. Stimuli were presented at two opposite horizontal orientations to investigate whether areas that are selective for tools also show greater activation when the tool's handle is directed towards the hand as opposed to away from it. Our results showed that, even after the low-level differences between tools and non-tools were controlled, tools evoked more activation in the tool network as well as in sensorimotor areas. The orientation of the tool handles did not mediate effects within these sensorimotor areas. In sum, when we passively view tools, even without an intent to act, functional associations are automatically evoked and these associations are not specific to a particular hand. •We examined human brain activation to real tools vs. scrambled versions.•Tool-selective activity was due to function not elongation nor real-world size.•Tool areas did not preferentially activate to different handle orientations.