The paper describes the improvements in the technique of radio-receiver testing that have been made at the National Physical Laboratory. The arrangements now used for this purpose having been described in detail in Part 1, a proposed specification of methods of testing broadcast receivers is given in Part 2, together with experimental results obtained on such receivers. This specification is examined as to its applicability to the relative comparison of widely different types of receivers, and a classification is suggested so as to diminish the number of tests which it is necessary to make in order to assign a figure of merit to a particular receiver.