To repeat what all students are supposed to know, even though probably no one has ever bothered to tell you: you are unlikely to be a real academic success if you limit yourself to a single book. If you don't understand something here, a fruitful approach is to read the parallel sections in a few other books. There is a decent chance that someone else's version of the same ideas will make more sense to you than mine do.
By no means is this the only textbook on statistical mechanics. Gibbs Elementary Principles in Statistical Mechanics [5] is the classic work. Elementary Principles ... is not really suitable as a textbook, because it treats almost no applications of the theory, is somewhat densely written, (necessarily) omits quantum topics, and has no homework problems. Any serious student of statistical mechanics must still at some point sit down and read Gibbs. Tolman's Statistical Mechanics [11] (1938) is another classic work, swollen by the inclusion of hundreds of pages on the then-new theory of the quantum. (Tolman's 1927 volume [12], now rather rare, is more readable.)
Let me note some of the other books which I view favorably. Of the more recent elementary texts, McQuarrie's Statistical Mechanics[13] is one of the few not to be dated in its treatment of post-World War II results on liquids and correlation functions. McQuarrie is usefully read as a graduate- level companion text to mine; his homework problems are plentiful and often excellent. Our treatments differ primarily with respect to the postulatory and pedagogical aspects of the theory, and with respect to the relative importance of classical and quantum statistical mechanics. I have tried not to duplicate his choice of modern applications. S. K. Ma's [14] text has had considerable billing for its attack on ensemble mechanics (the attack was somewhat exaggerated in the advertising material). Ma treats a variety of more recent results in an intuitive way. He also mixes results from the microcanonical and canonical ensembles in ways which may confuse some beginning students. Reichl's A Modern Course in Statistical Physics [15] emphasizes topics au courant as of date of publication among a vigorous part of the physics community. The discussions are thorough; perusal of the past decade's Physics Today will show how short the correlation time of au courancy is in modern physics.