Medical interviewing skills are integral to good medical care. In order to measure these skills an instrument has been developed, called the Maastricht History-Taking and Advice Checklist (MAAS). It has been studied with regard to interrater reliability and validity. In this study a revised version of the MAAS (MAAS-R), a check-list of concrete interview behaviour, has been investigated concerning feasibility and reliability for examination purposes. Audio-recordings were obtained of 24 doctors, each interviewing eight different standardized patients. The recordings were independently scored by three general practitioners trained in using the MAAS-R. The results of generalizability analysis, considering the influences of doctors, cases and raters, are encouraging. In order to overcome case-specificity feasible and reliable measurement can be accomplished with 8-10 cases in 2-21/2 hours of testing time, each case being scored by a different rater. Reliability improves considerably if assessment is restricted to basic interviewing skills.