Methods derived from information theory are used for the analysis of social mobility matrices. They permit us to give quantitative answers to such questions as: How does destination depend on origin in the occupational mobility table? This was possible through reformulating this and other questions in terms of the predictability-uncertainty dimension. The methods do not require any assumption regarding the relation between different occupational categories. At the same time they provide a finer analysis of mobility tables revealing aspects of mobility which would be masked by more gross methods.