Since the early 1980s, BBM has measured radio tuning using a three-week sweep, chosen from a pre-announced window of 5 or 6 weeks. This sweep technique is familiar to radio broadcasters, researchers and others. In the United States, since 1981 Arbitron has developed a measurement system that is not dependent on the three-week sweep system. Rather, the Arbitron system measures 12 week periods, four times a year in most major markets, for a forty-eight-week measurement - virtually continuous measurement over the year. In the BBM system, respondents for a Fall survey are recruited by telephone in the late summer and, if they agree to participate, are sent diaries for a sweep that takes place in mid-Fail. Generally, the waiting period between recruitment and survey weeks is 06-Aug weeks. Each week represents a random third of the recruited sample. Diarists are asked to keep a record of radio tuning for one week from 5 a.m. to 1 a.m. diaries are then returned to BBM for processing. Importantly, each week is treated as a separate survey and the three weeks are "projected' 1 to a population base and averaged to produced published data. This differs from the Arbitron system which, in a general sense, gathers all the diaries together into one database and projects it to the population base. This latter approach means there is less reliance on any individual weeks sample to produce an audience figure. BBM staff investigated the "extended" weeks of measurement approach as an alternative to the present system. The EWM system offers operational advantages, as well as addressing important methodological issues; which this paper addresses. Theoretically, there is a statistical superiority to a system whose base for the projection (i.e. Arbitron's) is measured over a longer period of time and is larger, (since weights are smaller and, theoretically, reliability is improved). Notwithstanding these obvious advantages, it was decided that BBM should test any new system prior to its introduction as a production standard. As a result of that test, it was concluded that, on a whole, the EWM approach was sufficiently superior to warrant changing to as soon as possible.