Contrast injection protocol is known to affect the estimation of kinetic parameters in functional CT. A novel method is proposed to maximize the precision of parameter estimates by modulating the contrast injection scheme. The method models the intravenous contrast bolus to be dispersed by a "patient function" to give rise to the arterial input function, which, in turn, carries the contrast agent to tissue leading to contrast enhancement. The covariance matrix analysis was applied to calculate the uncertainty of parameter estimates as the coefficients of variation (CV) in the adiabatic tissue homogeneity (ATH), two-compartment, and the modified Kety model in which tumor pathophysiology is modeled. An optimization scheme was used to determine the optimal injection protocol which would minimize the CV of a particular kinetic parameter. For clinical utility, a recommended injection protocol was suggested from a statistical analysis with the optimal injection protocols obtained from the first group of 12 patients with cervix cancer. The efficacy of the recommended injection protocol was tested with a second group of 12 patients. In addition, the robustness of the recommended injection protocol to longitudinal study has been investigated in the presence of variations in arterial input function and tumor pathophysiology. Based on the data of the second group of patients, and using the ATH model, the recommended biphasic injection of two boluses improves the precision in the estimation of blood flow and mean transit time (MTT), by 36.9% and 38.4%, respectively, compared to the standard uniphasic injection protocol in the CV. However, measurement of the permeability surface area product and extravascular extracellular space volume favors a single fast bolus of the same contrast amount. The two-compartment model and the modified Kety model also benefited from the single fast bolus. The effect of variation in the arterial input function and tumor pathophysiology on the applicability of the recommended injection was also investigated. Based on computer simulation for a range of variations in the arterial input function and pathophysiology, the recommended biphasic injection was found to improve the precision in blood flow and MTT estimates by 31.4% and 36.5% on average, respectively, compared to the uniphasic injection.