I present a cumulative case for the thesis that we only know propositions that are certain for us. I argue that this thesis can easily explain the truth of eight plausible claims about knowledge:

There is a qualitative difference between knowledge and non-knowledge.

Knowledge is valuable in a way that non-knowledge is not.

Subjects in Gettier cases do not have knowledge.

If S knows that P, P is part of S’s evidence.

If S knows that P, ~P is epistemically impossible for S.

If S knows that P, S can rationally act as if P.

If S knows that P, S can rationally stop inquiring whether P.

If S knows each of {P1, P2, … P n }, and competently deduces Q from these propositions, S knows that Q.

I then argue that the skeptical costs of this thesis are outweighed by its explanatory power.