Preclinical experiments are carried out with approximately 20-30 microm wide, approximately 10 mm high parallel microbeams of hard, broad-"white"-spectrum x rays (approximately 50-600 keV) to investigate microbeam radiation therapy (MRT) of brain tumors in infants for whom other kinds of radiotherapy are inadequate and/or unsafe. Novel physical microdosimetry (implemented with MOSFET chips in the "edge-on" mode) and Monte Carlo computer-simulated dosimetry are described here for selected points in the peak and valley regions of a microbeam-irradiated tissue-equivalent phantom. Such microbeam irradiation causes minimal damage to normal tissues, possible because of rapid repair of their microscopic lesions. Radiation damage from an array of parallel microbeams tends to correlate with the range of peak-valley dose ratios (PVDR). This paper summarizes comparisons of our dosimetric MOSFET measurements with Monte Carlo calculations. Peak doses at depths <22 mm are 18% less than Monte Carlo values, whereas those depths >22 mm and valley doses at all depths investigated (2 mm-62 mm) are within 2-13% of the Monte Carlo values. These results lend credence to the use of MOSFET detector systems in edge-on mode for microplanar irradiation dosimetry.