The definition and objectives of robustness testing are given and the essential features of the methodology which can be applied using a multivariate approach in liquid chromatography and capillary electrophoresis are described. Guidelines are given for the different steps which are involved in using screening and response surface designs. It is shown that screening designs may be sufficient to set the method limits but that response surface designs are of major interest in method transfer because they give a comprehensive picture of the behaviour and limitations of the method.