The controlled Markov chains (CMC) approach to software testing models the software under test as a controlled Markov chain and treats it as a controlled object. It treats software testing strategy as the corresponding controller. Our previous work has shown that the CMC approach is applicable to dealing with the software testing problem for optimal software reliability growth. In this paper we show that software reliability assessment can be treated as an adaptive control problem and the CMC approach is also applicable to dealing with the optimal software testing problem for software reliability assessment. The work presented in this paper is a contribution to the new area of software cybernetics that explores the interplay between software and control.