Neuropathology is often mediated by interactions between neurons and glia that cannot be modeled by monocultures. However, cocultures are difficult to use and analyze for high-content screening. Here, we perform compound screening using primary neuron-glia cultures to model inflammatory neurodegeneration, live-cell stains, and automated classification of neurons, astrocytes or microglia using open-source software. Out of 227 compounds with known bioactivities, 29 protected against lipopolysaccharide-induced neuronal loss, including drugs affecting adrenergic, steroid, inflammatory and MAP kinase signaling. The screen also identified physiological compounds, such as noradrenaline and progesterone, that protected and identified neurotoxic compounds, such as a TLR7 agonist, that induced microglial proliferation. Most compounds used here have not been tested in a neuron-glia coculture neurodegeneration assay previously. Thus, combining a complex cellular disease model with high-content screening of known compounds and automated image analysis allows identification of important biology, as well as potential targets and drugs for treatment.