OK finally I have read it over (at Baltimore MD), though pretty roughly. In recent months, my majority brain cycles were on this book. And compared with other CS books I read before, first, it is on TCS not on engineering, and this one is the most abstract and brain-demanding one - but actually it's said to be the most accessible book on the topic. I still remember in my last semester in grad school I borrowed Hopcroft and tried to read it. With several useless struggling I failed to understand anything further than on first several pages.
Not sure how much I've digested and remembered about this book (considering my experience of it: 5 months in spare time), but what i have learned is still enormous: the way Sipser conducted his proofs, and in future where I can refer to in case of any TCS related topics appeared in my work - though Pr <= 2^(-n) :P