The basic contributions to the concept of the many-worlds interpretation of quantum mechanics are analysed. It is found necessary to divide them into two classes corresponding to (a) the relative-states interpretation (RSI), which avoids problems of measurement, but is found generally unconvincing, and (b) the many-worlds interpretation (MWI), which is more comprehensible, but has all the problems of the von Neumann scheme. The EPR problem is tackled using each interpretation. The RSI finds no problem, while the MWI meets the same difficulties as conventional interpretations.