One of the most delicate problems in the use of control charts to maintain current control of a process is the choice of sample size and statistical errors. It is, in fact, commonly accepted that a sampling procedure cannot be defined unless a few targets are fixed. On the other hand, it is impossible to eliminate all degrees of arbitrariness in the choice of the quantities which characterize a sampling plan. Therefore, in the practice of control charts, very low values of the risk of false 'out of control' signals are usually fixed. In fact, it is generally argued that the consequent higher risk of not detecting a shift in the process, and the resulting increase of nonconformities in the productwhich can be reduced in any case by increasing the size of the samples are less costly than arresting a process under control. The procedure requires some of the quantities which define the sampling desing of the charts and determine the remaining quantities to be arbitrarily predetermined. In order to reduce as much as possible the elements of arbitrariness present in the construction of a control chart several models have been proposed that take into account the costs one has to sustairras a consequence of the choices made for that purpose. The economic design of control charts can be therefore considered a classic topic, whose foundations in rigorous terms after early pioneer works by some other authors can be traced back to Duncan (1956), who studied the problem with reference to ~'-charts. Other approaches have been introduced over more than 25 years, with the development of models both for charts by variables and attributes (for the latter see, for example, Chiu, 1975). An extensive review of different models can be found in Montgomery (1985). Mostly, the models proposed are of the single assignable, or multiple assignable cause type (Duncan, 1956 and 1971; Fetter, 1967; Knappenberger and