The aim of the paper is to examine Tichý’s understanding of the term “assumption”. We show that Tichý distinguishes two approaches to inference: the one-dimensional view that treats inferences as a sequences of logical rules or axioms as well as hypotheses and their logical consequences; and the two-dimensional view specifying inference as a derivation of one entailment from (the set of) another entailment(s). It is claimed that Tichý is right in his critique of Meinong’s concept of assumption as ‘assertion without conviction’. Nevertheless, Tichý – in addition to his logical concept of assumption – uses, though unreflectively, also the epistemic concept of assumption. Henceforth, we claim that accepting Tichý’s rejection of the epistemically hypothetical assumptions we couldn’t use logic as an instrument for empirical knowledge enhancement. We believe, to the contrary, that the epistemic assumptions may become a basis for derivations and knowledge enhancement, even though they do not represent necessary truths.