A number of designs, including filterpress-type cells, de Nora's Seachlor cell, Capenhurst's Chemelec cell, Nalco's packed bed cell and Akzo's fluidized-bed cell, are described briefly. An attempt is made to determine the reasons why these particular designs are finding application in industry. It is shown that each performs a necessary function with the minimum of complexity. Thus filterpress-type cells are relatively compact, and provided current densities are not too low will be the first choice where the electrodes do not change dimensions during the reaction. The Seachlor cell, by virtue of its particular bipolar electrode arrangement, controls pH and therefore minimizes any deposition of magnesium hydroxide during the production of hypochlorite from seawater. The Chemelec cell increases limiting currents for metal deposition in a relatively simple manner while the Nalco cell solves the problem of the participation of its lead anodes in the manufacture of lead tetraalkyls. The fluidized-bed electrode, with its high space time yield and the production of very pure metals, makes up for its more complex character. The paper shows why some of the alternatives to the above designs are less attractive to industry. Having decided upon a particular design, the paper then discusses some erroneous ideas on optimization encountered occasionally in the literature. An example is given where the `optimization' of the cell in isolation of its associated plant can result in a very misleading `optimum'. It is demonstrated that failure to consider the process as a whole, right from its earliest stages of development in the research laboratory, can lead to severe problems when engineering of the process is eventually required.