In an effort to explain how magnetic storms can cause strong enhancements in outer belt MeV electrons, we have studied the substorm-associated acceleration of energetic (tens of keV) plasma sheet electrons and their injection into the outer-trapped region of the magnetosphere. The study is based on tracing test particles in three-dimensional MHD simulations of substorm dipolarization. The simulation-based electric and magnetic fields [Birn and Hesse, 1996> exhibit a strong earthward collapse around local midnight, leading to magnetic field dipolarization along with a corresponding induction electric field. The test particle traces show that tens of keV plasma sheet electrons can be transported from about x≈-20 RE to x≈-10 RE and can gain about a factor of 10 in energy. If these particles are further transported inward to L~6 while conserving the first adiabatic invariant, they will have energies of an MeV or more. In the substorm acceleration process the dominant energy gain occurs during the earthward radial transport by E¿B drift in the strong dipolarization region. In our calculations the electrons that gain the most energy are ones that circle a local maximum in the equatorial magnetic field strength and thus spend a relatively long time in the region of collapsing field. The first adiabatic invariant is broken at the beginning of the particle trajectory, near the neutral line, where the magnetic curvature radius can be comparable to the particle gyroradius. The second adiabatic invariant is also broken later in the process, when particles can be temporarily trapped in off-equatorial magnetic minima associated with MHD waves on the dipolarizing field lines. Estimation of the number of accelerated plasma sheet electrons indicates that the Birn-Hesse substorm, which is not particularly large, produces only ~2% of the number of MeV electrons observed in a typical poststorm outer belt electron enhancement. A series of substorms, some of them large, might produce a large enough enhancement, but it should be noted that this is an order of magnitude estimate, and it is rather sensitive to plasma sheet and substorm parameters. ¿ 2000 American Geophysical Union