The unobservability thesis (UT) states that the mental states of other people are unobservable. Both defenders and critics of UT seem to assume that UT has important implications for the mindreading debate. Roughly, the former argue that because UT is true, mindreaders need to infer the mental states of others, while the latter maintain that the falsity of UT makes mindreading inferences redundant. I argue, however, that it is unclear what ‘unobservability’ means in this context. I outline two possible lines of interpretation of UT, and argue that on one of these, UT has no obvious implications for the mindreading debate. On the other line of interpretation, UT may matter to the mindreading debate, in particular if we think of it as a thesis about the possible contents of perceptual experience. The upshot is that those who believe UT has implications for the mindreading debate need to be more specific about how they understand the thesis.