My dissertation treats theories of evidence. The main questions addressed therein are, what questions does a theory of evidence aim to answer? are the popular extant theories of evidence satisfactory? what other possible competing theories are worthy of consideration? and do the intuitions about the evidence people have in so-called ``bad cases' (evil-demon, BIV) give us conclusive reason to reject otherwise attractive theories of evidence? Much recent epistemology has proceeded without explicitly considering these questions. This dissertation offers a critical examination of recent theories of evidence, and a constructive proposal for a new one, thus filling a significant gap in our philosophical understanding. The dissertation divides into two parts. In the first part, I critically examine some other theories of evidence. First, I examine some members of a family of theories of evidence that are committed to Propositionalism, which is the thesis that all evidence consists in propositions. Second, I examine a theory I call Rochester Evidentialism, which is the theory of evidence developed and defended by Earl Conee and Richard Feldman for the past twenty-five years. In the second part, I articulate and defend a new theory of evidence that I call Factualism. A fundamental commitment of Factualism is that the fundamental evidence role is that of indicating some truth to someone. This conception of the fundamental evidential role helps us derive an ontology of evidence according to which all evidence consists in epistemically-usable facts, where facts here are not understood a true propositions, but as that which true propositions represent. Intuitively, what makes a fact epistemically usable for someone is that she possess certain cognitive capacities which permit her to exploit the fact in her cognition. The notion of evidential support is cashed out in terms of epistemic usability, specific fact-types, and partial dependence. Put roughly, a fact e evidentially supports p for S iff e is epistemically usable for S and either most tokens of e's specific fact-type partly depend on a token of the specific fact-type of which p represents a token, or most tokens of the specific fact-type of which p represents a token depend on some token of e's specific fact-type. Evidence possession is defined in terms of what a subject knows, and a kind of awareness that I call t-awareness, which takes as its objects both internal and external facts, and does not entail the having of any beliefs. I then define a subjects total evidence at a time as including all and only those facts which she either knows, or is t-aware of at that time.