In this paper, I will show that Sherlock Holmes was a good logician, according to the standard of the 19th century, both in his character and knowledge (sections 2 and 3). Holmes, in all probability, knew William Stanley Jevons’ clarification of deductive reasoning in terms of “logical alphabets” (section 4). And in view of his use of “analytic-synthetic” distinction and “analytic reasoning,” I will argue that Holmes knew rather well philosophy too, as far as logic and methodology are concerned (section 5). Further, I have argued that Holmes introduced new twists (presumably, following Jevons) into analytic reasoning: application to reasoning as regards causal sequences, and probabilistic elimination of hypotheses (sections 6 and 7). Also, in this context, I will clarify the significance of Holmes’ metaphor of the “little attic”: without fine assortment in your brain, it is hard to devise promising hypotheses (section 8). Finally, presenting a simple model of probabilistic inference, which became prevalent in the 19th century (section 9), I claim that the essence of Holmes’ reasoning consists of probabilistic inferences, “balance probabilities and choose the most likely,” which is nothing but probabilistic elimination of hypotheses in the light of evidence. I also argue that my claim fits in well with the text of Holmes stories (section 10).