Immanuel Kant considers hypotheses as decisive for the assessment and development of scientific enquiry. Hypotheses guide the acquisition of data and the constructing of experiments. Yet, hypotheses are not reliable sources of knowledge since they can be mistaken for other forms of assumptions that may be unproductive for scientific discovery. Hypotheses are not reliable for two main reasons: either they prevent one starting a proper enquiry or lead one astray from it. The involvement of imagination within the process of hypothesis formation appears to be the source of mistakes since possible causes of observed phenomena may be merely feigned. This paper argues that Kant theorises a method of forming hypotheses that distinguishes hypotheses from other kinds of assumptions, especially fancies, at the moment of their formation. This method warrants that hypotheses are not feigned, according to Isaac Newton’s dictum, while preserving the fallible status of hypotheses. A hypothesis can turn out to be false on empirical testing, but this does not imply it was a fancy.