The use of correlated coefficients of heat transfer from the experimental data has been the universal choice to analyze complex heat transfer processes. Unfortunately, large errors often occur in the application of such coefficients to complex phenomena such as those occurring in compact fin-tube heat exchangers. Basic issues responsible for this difficulty are identified and articulated. An alternative approach of correlating the experimental heat transfer data by the application of artificial neural network (ANN) is presented, and this new approach is shown to possess several inherent advantages not enjoyed by the conventional approach of using correlated heat transfer coefficients, as demonstrated by examples dealing with two fin-tube heat exchangers with different complexities. Also mentioned are further extensions of the ANN approach to the dynamic modeling of heat exchangers and their control.