Reaching the full potential of magnetic resonance imaging (MRI)-positron emission tomography (PET) dual modality systems requires new methodologies in quantitative image analyses. In this study, methods are proposed to convert an arterial input function (AIF) derived from gadolinium-diethylenetriaminepentaacetic acid (Gd-DTPA) in MRI, into a 18F-fluorodeoxyglucose (18F-FDG) AIF in PET, and vice versa. The AIFs from both modalities were obtained from manual blood sampling in a F98-Fisher glioblastoma rat model. They were well fitted by a convolution of a rectangular function with a biexponential clearance function. The parameters of the biexponential AIF model were found statistically different between MRI and PET. Pharmacokinetic MRI parameters such as the volume transfer constant (Ktrans), the extravascular–extracellular volume fraction (νe), and the blood volume fraction (νp) calculated with the Gd-DTPA AIF and the Gd-DTPA AIF converted from 18F-FDG AIF normalized with or without blood sample were not statistically different. Similarly, the tumor metabolic rates of glucose (TMRGlc) calculated with 18F-FDG AIF and with 18F-FDG AIF obtained from Gd-DTPA AIF were also found not statistically different. In conclusion, only one accurate AIF would be needed for dual MRI-PET pharmacokinetic modeling in small animal models. Magn Reson Med, 2013. © 2012 Wiley Periodicals, Inc.