Microbeam radiation therapy (MRT) is a currently experimental method of radiotherapy which is mediated by an array of parallel microbeams of synchrotron-wiggler-generated x-rays. Suitably selected, nominally supralethal doses of x-rays delivered to parallel microslices of tumor-bearing tissues in rats can be either palliative or curative while causing little or no serious damage to contiguous normal tissues. Although the pathogenesis of MRT-mediated tumor regression is not understood, as in all radiotherapy such understanding will be based ultimately on our understanding of the relationships among the following three factors: (1) microdosimetry, (2) damage to normal tissues, and (3) therapeutic efficacy. Although physical microdosimetry is feasible, published information on MRT microdosimetry to date is computational. This report describes Monte Carlo-based computational MRT microdosimetry using photon and/or electron scattering and photoionization cross-section data in the 1 eV through 100 GeV range distributed publicly by the U.S. Lawrence Livermore National Laboratory (LLNL) in the 1990s. These are compared with Monte Carlo-based microdosimetric computations using a code and physical data available in the 1980s. With the aim of using the PSI-version of GEANT Monte Carlo code for future macro- and micro/nano-dosimetric studies of Microbeam Radiation Therapy (MRT) a comparison of this code is made with the INHOM(EGS4) (version 1990), Dilmanian-CPE and Persliden-CPE Monte Carlo photon-electron codes (both version 1990) with which the absorbed dose distributions were calculated in 1990 and 1991 considering, (a) a single cylindrical microbeam, (b) multiple cylindrical microbeams in an orthogonal square bundle, and (c) multiple planar microbeams. It is shown that the PSI-version of GEANT can potentially deliver more accurate results (a) using presently the most advanced atomic data, and especially (b) employing "Single-collision" electron transport instead of only the "Condensed-history" electron transport as in code INHOM(EGS4). In contrast Dilmanian-CPE and Persliden-CPE codes deposit the electron energy locally instead of transporting it to the correct position.