The original data of 109 pairs of RR and QT intervals published by Dr. Bazett in 1920 were subjected to a computerized optimization of the QT/RR relationship. Four generic formulae expressing QT interval as alpha + beta RR mu, alpha + beta e mu RR, alpha + beta e(1/mu RR), and alpha + beta ln mu(10 x RR) were used, and for each formula, optimum combinations parameters alpha, beta, and mu were established that lead to minimum differences between the actual QT interval intervals and values predicted by the model. The results show that independent of the generic formula used, parameters of the model can be selected that provide a close fit of the measured values. Compared to the optimum models obtained in this way, the optimum fit of the Bazett model, QT = beta RR1/2, overestimates the QT intervals for slow heart rates. In addition to demonstrating the procedure of obtaining an optimum QT/RR model for a given data set, the study suggests that critical understanding of the principal limitations of every QT/RR model is more important than its mathematical form.