PET-CT is probably the ultimate tool for accurate tumor imaging and 3-dimensional in vivo predictive assay of radiation sensitivity. By imaging the tumor twice during the early course of therapy, it should be possible to quantify both the tumor responsiveness to therapy and the rate of loss of functional tumor cells using the presently derived equations. This new information is ideal for use together with biologically based therapy optimization and makes it possible accurately to quantitate the dose-response relation, at least for the bulk of the tumor cells. Since the tumor responsiveness is available after about one and a half weeks of therapy, the information is also ideal for use with adaptive therapy where all forms of deviations from the original treatment plan can be accurately corrected for since they generally influence the still functional, but mainly doomed tumor cell compartment. Thus, uncertainties such as: 1) the geometric misalignment of the therapeutic beam with the tumor, 2) deviations of the delivered dose distribution from the planned delivery whether due to 3) an erroneous treatment planning algorithm or 4) treatment equipment uncertainties and 5) deviations in the anticipated responsiveness of the tumor of the patient based on historical response data, can all be taken into account. Fortunately, when a larger tumor cell compartment than expected is seen an increased dose during the remainder of the treatment should always be delivered independently on whichever combination of the above deviations was the true reason. With high-energy photon and hadron therapy it is even possible to image the integral dose delivery in vivo during or after a treatment using PET-CT imaging. The high-energy photons above about 20 MeV produce positron emitters through photonuclear reactions in tissue which are proportional to the photon fluence and thus approximately also to the absorbed dose. Light ion beams, the ultimate radiation modality with regard to physical and biological selectivity, instead produce PET emitters through direct nuclear interactions in tissue, but can also be used as radioactive beams consisting of intrinsic PET emitters such as 8B, 11C, 13N and 15O. These radioactive beams allow more accurate imaging of the Bragg peak distribution and thus indirectly the absorbed dose. The most universal feedback for adaptive radiation therapy would then be to use the measured image of mean dose delivery during the early part of the treatment while revising the treatment plan based on the initially planned dose distribution and the radiation responsiveness of the tumor as seen after the first week or two of therapy. By this so-called BIO-ART approach (Biologically Optimized 3D in vivo predictive Assay-based Radiation Therapy) radiation therapy optimization may become an almost exact science, where the patient's true individual radiation response, considering hypoxia and general radiation resistance as well as possible dose delivery and planning errors, is taken into account.