The concept of tailored testing for individuals is applied to the construction of tests for special groups and extended to apply to item content as well as item difficulty. It is suggested that evaluators may decide to construct tests on the basis of a unique combination of items drawn from an item bank to fit the need of a particular group. At the same time they may be interested in relating achievements obtained on such tests to some population norms. In order to do this it is necessary to estimate the population mean and standard deviation of each test composed of a unique set of items which, in the past, had not been administered as a single test. This requires finding a way to estimate population standard deviations. The present paper explores the possibility of using two techniques for estimating this parameter; one is Horst's formula for estimating standard deviation and the second one is based on the estimated interitem correlation matrix. In order to examine the adequacy of these estimation formulae, 30 sets of items were selected from the item bank, and for each of these sets the estimated values were compared with the observed ones. (Author/CTM)