In the periodic testing environment associated with conventional paper-and-pencil tests, the frequency with which items are seen by test-takers is tightly controlled in advance of testing by policies that regulate both the reuse of test forms and the frequency with which candidates may take the test. In the continuous testing environment associated with more novel testing paradigms such as computerized adaptive testing (CAT), the computer itself can be used to control the frequency with which items are administered. This paper discusses previous methods of controlling item security in the continuous adaptive testing environment and presents a new method that overcomes some (but not all) of the disadvantages of previous attempts. The new method rethinks the use of the exposure control parameters in selecting each item to be administered. An extensive sample with this new method and a particular adaptive testing algorithm illustrates how concerns about test efficiency, parallelism, and security can be balanced. The target population was estimated using the method of R. J. Mislevy (1984) and a sample of over 5,000 real test-takers who took a linear 50-item paper-and-pencil analytical reasoning test to which the adaptive tests were designed to be parallel. (Contains 2 figures and 19 references.) (Author/SLD)