[en] This thesis investigates the validity of inverse algorithms and delivery techniques in dynamic intensity modulated radiation therapy (IMRT) with the view toward clinical implementation. A dosimetric verification technique used to monitor the dynamic beam delivery was introduced with design considerations and clinical evaluation of various quality assurance (QA) phantoms, facilitating the measurement of dynamic IMRT dose distributions and conversion of photon fluence to machine deliverable monitor units. Benchmark tests using thermoluminescent dosimeters (TLDs), solid state detectors, and films were carried out in realistic clinical examples. This was to demonstrate the precision and degree of accuracy of the proposed techniques both in vitro and in vivo systems using canine subjects with paraspinal tumors. The results of the irradiation of canine subjects indicated that the goal of using IMRT to irradiate target volumes with sparing critical structures was met. However, the measure of physiological response found, was incapable of providing a more accurate measure of the dose delivered. Quantitative analysis on the influence of systematic and random field perturbations in highly conformal fields with emphasis on the dosimetric outcome is also presented in this thesis. Two sources of uncertainties deemed important in dynamic therapy of photon beams were studied. These were caused by inaccuracies in velocity and in position of both multileaf collimator (MLC) and backup diaphragms. This was attributed to some degree to the tolerance levels or mechanical constraints of beam defining parameters. These errors were simulated using a Gaussian function with standard deviation of ±1.0 mm to assess changes in dose distribution. Based on the results of this study, QA procedures which are unique to the delivery of dynamic beams were proposed