Inverse methods are often used in heat transfer analysis to determine parameters which are difficult or impossible to measure directly, from carefully selected experimental measurements which can. be more easily carried out. In this type of procedure, there are often a large number of boundary conditions which can affect the outcome. Often it is impractical, or even impossible, to accurately measure all influential boundary conditions, and thus estimates are employed; However, the question naturally arises, how sensitive are the results of the inverse analysis to the assumed boundary conditions? A statistical method, employing an orthogonal array matrix designed experiment, which was used to estimate the effect of the uncertain boundary conditions on the results of the analysis, is described in detail and the results are given. (C) 1997 Elsevier Science Ltd.