The mechanism of intercalation into DNA double helices by ethidium has been analyzed by temperature-jump relaxation and stopped-flow measurements using fluorescence detection. Artifacts due to field- or flow-induced alignment have been eliminated by measurements under magic angle conditions; the theoretical basis for suppression of orientation effects resulting from external forces is given for the case of fluorescence measurements. Excluded site effects have been avoided by restriction to low degrees of binding. The temperature-jump relaxation observed for ethidium binding to DNA could be described by single exponentials under most conditions. The reciprocal time constants increased linearly with the DNA concentration, leading to association rate constants of 2.7 X 10(6) M-1 s-1 at 12-degrees-C. These rate constants are virtually independent of the DNA chain length for samples with 200, 500, 4228, and 30 000 base pairs, showing that the rate is controlled by reaction and not by a diffusive process. At high DNA concentrations around 200 muM, an additional relaxation effect with an amplitude opposite to the main one is observed which is probably due to some conformational change of the DNA-ethidium complex. The results obtained by stopped-flow measurements are consistent with those from T-jump measurements, but owing to higher amplitudes and better signal to noise ratios, the stopped-flow data clearly require two exponentials for satisfactory representation. The reciprocal time constants for both processes increase linearly with the DNA concentration. The simplest mechanism consistent with this result involves parallel formation of two different complexes with a direct transfer of ethidium between the binding sites. The experimental data for a synthetic DNA with one type of base pairs, poly[d(A-T)], are very similar to those for natural DNA; thus, the relatively complex reaction mechanism of intercalation is not due to base-pair heterogeneity. Apparently the two complexes are formed by intercalation of ethidium between the base pairs from the directions of the major and the minor grooves of the double helix. Our results demonstrate that the Joule temperature-jump technique does not introduce artifacts due to field-induced processes, when this technique is used with appropriate care.