Cerebral metabolic rate of oxygen (CMRO(2)), oxygen extraction fraction (OEF) and cerebral blood flow (CBF) images can be quantified using positron emission tomography (PET) by administrating (15)O-labelled water (H(15)(2)O) and oxygen ((15)O(2)). Conventionally, those images are measured with separate scans for three tracers C(15)O for CBV, H(15)(2)O for CBF and (15)O(2) for CMRO(2), and there are additional waiting times between the scans in order to minimize the influence of the radioactivity from the previous tracers, which results in a relatively long study period. We have proposed a dual tracer autoradiographic (DARG) approach (Kudomi et al 2005), which enabled us to measure CBF, OEF and CMRO(2) rapidly by sequentially administrating H(15)(2)O and (15)O(2) within a short time. Because quantitative CBF and CMRO(2) values are sensitive to arterial input function, it is necessary to obtain accurate input function and a drawback of this approach is to require separation of the measured arterial blood time-activity curve (TAC) into pure water and oxygen input functions under the existence of residual radioactivity from the first injected tracer. For this separation, frequent manual sampling was required. The present paper describes two calculation methods: namely a linear and a model-based method, to separate the measured arterial TAC into its water and oxygen components. In order to validate these methods, we first generated a blood TAC for the DARG approach by combining the water and oxygen input functions obtained in a series of PET studies on normal human subjects. The combined data were then separated into water and oxygen components by the present methods. CBF and CMRO(2) were calculated using those separated input functions and tissue TAC. The quantitative accuracy in the CBF and CMRO(2) values by the DARG approach did not exceed the acceptable range, i.e., errors in those values were within 5%, when the area under the curve in the input function of the second tracer was larger than half of the first one. Bias and deviation in those values were also compatible to that of the conventional method, when noise was imposed on the arterial TAC. We concluded that the present calculation based methods could be of use for quantitatively calculating CBF and CMRO(2) with the DARG approach.