Abstract: Several years ago, Theorell and Czerlinski conducted experiments on the system of horse liver alcohol dehydrogenase, reduced nicotinamide adenine dinucleotide and imidazole, using the first version of the temperature jump apparatus with detection of changes in fluorescence. These early experiments were repeated with improved instrumentation and confirmed the early experiments in general terms. However, the improved detection system allowed to measure a slight concentration dependence of the relaxation time of around 3 ms. Furthermore, the chemical relaxation time was smaller than the one determined earlier (by factor 2). The data were evaluated much more rigorously than before, allowing an appropriate interpretation of the results. The observed relaxation time is largely due to rate constants in an interconversion of ternary complexes, which are faster than three (of the four) dissociation rate constants, determined previously by Theorell and McKinley-McKee.1,2 This fact contributed to earlier difficulties of finding any concentration dependence. However, the binding of imidazole to the binary enzyme-coenzyme complex can be made to couple kinetically into the interconversion rate of the two ternary complexes. The observed signal derives largely from the ternary complex(es). A substantial fluorescence signal change is associated with the observed relaxation process, suggesting a relocation of the imidazole in reference to the nicotinamide moiety of the bound coenzyme. Nine models are considered with two types of coupling of pre-equilibria (none-all). Quantitative evaluations favor the model with two ternary complexes connected by an interconversion outside the four-step (bimolecular) cycle. The ternary complex outside the cycle has much higher fluorescence yield than the one inside. The interconversion equilibrium is near unity for imidazole. If it would be shifted very much to the side of the “dead-end” complex (as in isobutyramide?!), stimulating action could not take place.