An improved method of analysis is presented for the interpretation of fouling in shell and tube heat exchangers. The method reveals weaknesses that are inherent in commonly used but simpler approaches to the interpretation of plant data. It is shown, for example, that by neglecting the effect of by-passing and the effect of heat transfer coefficient variation through an exchanger, it is possible to become misled into believing that a heat exchanger may be fouled when actually it is clean. The improved method of analysis shows that the corresponding 'apparent fouling resistance' of a clean heat exchanger is higher the lower its temperature effectiveness. The improved method is applied firstly to a hypothetical heat exchanger located in a train of clean exchangers and secondly to the shell and tube exchangers in an oil refinery crude preheat train.