A methodology that is adequate for a generalized description of the interactive problem-solving behavior of client and clinician must accommodate all potential types of clinical issues and client-clinician relationships. It must also be applicable to the behavior of the client as well as that of the clinician. Furthermore, behavior must be analyzable in terms of the phase of problemsolving logic that applies at the time the behavior is observed. The development, specifications, structural factors and procedures of administration of an observational methodology that is based on a 7-phase paradigm of verbal problem-solving behavior are described. Acceptable levels of consistency of coding are achievable using the methodology. Coder agreement over all phase categories was 84.8 per cent. Results of 2-way ANOVA procedures showed that the type of clinical encounter observed affected the proportion of data coded in each of the problem-solving phase categories whereas, on the whole, the identity of the coder did not. The methodology may be articulated with other methods for the study of primary care and has both research and instructional applications.