Background: Accurate knowledge of fission fragment yields is an essential ingredient of numerous applications ranging from the formation of elements in the r-process to fuel cycle optimization for nuclear energy. The need for a predictive theory applicable where no data is available, together with the variety of potential applications, is an incentive to develop a fully microscopic approach to fission dynamics.Purpose: In this work, we calculate the pre-neutron emission charge and mass distributions of the fission fragments formed in the neutron-induced fission of 239 Pu using a microscopic method based on nuclear density functional theory (DFT).Methods: Our theoretical framework is the nuclear energy density functional (EDF) method, where large amplitude collective motion is treated adiabatically using the time dependent generator coordinate method (TDGCM) under the Gaussian overlap approximation (GOA). In practice, the TDGCM is implemented in two steps. First, a series of constrained EDF calculations map the configuration and potential energy landscape of the fissioning system for a small set of collective variables (in this work, the axial quadrupole and octupole moments of the nucleus). Then, nuclear dynamics is modeled by propagating a collective wave packet on the potential energy surface. Fission fragment distributions are extracted from the flux of the collective wave packet through the scission line.Results: We find that the main characteristics of the fission charge and mass distributions can be well reproduced by existing energy functionals even in two-dimensional collective spaces. Theory and experiment agree typically within 2 mass units for the position of the asymmetric peak. As expected, calculations are sensitive to the structure of the initial state and the prescription for the collective inertia. We emphasize that results are also sensitive to the continuity of the collective landscape near scission.Conclusions: Our analysis confirms that the adiabatic approximation provides an effective scheme to compute fission fragment yields. It also suggests that, at least in the framework of nuclear DFT, three-dimensional collective spaces may be a prerequisite to reach 10% accuracy in predicting pre-neutron emission fission fragment yields.