Sinai's walk $(S_n, \; n\ge 0)$ is a recurrent one-dimensional nearest-neighbour random walk in random environment, and is reputed for its exotic slow movement : $S_n \approx (\log n)^2$, as $n$ goes to infinity. The present article summarizes the approach via stochastic calculus in the study of Sinai's walk. The main tool is the Ray–Knight theorem which describes the local time process of Brownian motion stopped at some special random times. The method is very powerful. For example, it allows to establish all the possible Lévy es for Sinai's walk and to determine the escape rate of favourite sites. It is interesting to mention that the latter problem remains open for the usual random walk. A number of unanswered questions, which concern various asymptotic properties of Sinai's walk, are listed at the end of the article.