Two hypotheses have been presented to explain the grossly biphasic oxygen release kinetics observed when hemoglobins are studied with the oxygen pulse technique [Gibson (1973) Proc. Nat. Acad. Sci. USA 70, 1-4]. Hypothesis I suggests that the two phases result from cooperativity, with the fast phase being oxygen release from the low affinity (T) state and the slow phase being oxygen release from molecules that have switched to the high affinity (R) state. Hypothesis II suggests that the biphasic curves are due to a large (factor of 20-30) difference in oxygen release from the two types of subunits within deoxyhemoglobin. In order to experimentally discriminate between these two hypotheses, we reinvestigated the oxygen pulse reaction for hemoglobin Kansas (alpha2 beta2 102 Asn leads to Thr) in the absence and presence of inositol hexaphosphate, since recent high resolution nuclear magnetic resonance studies have shown that this allosteric cofactor stabilizes hemoglobin Kansas in T even when fully liganded [Ogawa, Mayer, and Shulman (1972) Biochem. Biophys. Res. Commun. 49, 1485-1491]. The results of these studies clearly favor hypothesis I over hypothesis II as being the correct interpretation for the oxygen pulse results. However, we have found evidence that suggests that oxygen release and binding in T are surprisingly faster than previously observed. Furthermore, within T, there is some spectral and kinetic heterogeneity for oxygen release from adult hemoglobin and hemoglobin Kansas. The magnitude of this kinetic heterogeneity in T appears to be about the same as that seen in the high affinity, R, state. The exchange of hypothesis II for hypothesis I more strongly favors views of cooperative oxygen binding involving both types of subunits, as required if the allosteric model of Monod, Wyman, and Changeux [(1965) J. Mol. Biol. 12, 88-118] is considered operative.