The dynamics of the binding reaction of ANS to native and partly folded (molten globule) tuna and horse apomyoglobins has been investigated by fluorescence correlation spectroscopy and frequency domain fluorometry. The reaction rate has been measured as a function of apomyoglobin and ANS concentrations, pH, and temperature. Examination of the autocorrelation functions shows that the reaction rate is fast enough to be observed in tuna apomyoglobin, whereas the reaction rate in horse apomyoglobin is on the same time scale as diffusion through the volume or longer. Specifically, for tuna apomyoglobin at pH 7 and room temperature the on rate is 2200 microM(-1) s(-1) and the off rate is 5900 s(-1), in comparison with k(on) = 640 microM(-1) s(-1) and k(off) = 560 s(-1) for horse myoglobin as measured previously. The independence of the reaction rate from the ANS concentration indicates that the reaction rate is dominated by the off rate. The temperature dependence of the on-rate shows that this rate is diffusion limited. The temperature dependence of the off rates analyzed by Arrhenius and Ferry models indicates that the off rate depends on the dynamics of the protein. The differences between horse and tuna apomyoglobins in the ANS binding rate can be explained in terms of the three-dimensional apoprotein structures obtained by energy minimization after heme removal starting from crystallographic coordinates. The comparison of the calculated apomyoglobin surfaces shows a 15% smaller cavity for tuna apomyoglobin. Furthermore, a negative charge (D44) is present in the heme cavity of tuna apomyoglobin that could decrease the strength of ANS binding. At pH 5 the fluorescence lifetime distribution of ANS-apomyoglobin is bimodal, suggesting the presence of an additional binding site in the protein. The binding rates determined by FCS under these conditions show that the protein is either in the open configuration or is more flexible, making it much easier to bind. At pH 3, the protein is in a partially denatured state with multiple potential binding sites for ANS molecule, and the interpretation of the autocorrelation function is not possible by simple models. This conclusion is consistent with the broad distribution of ANS fluorescence lifetimes observed in frequency domain measurements.