Treatment with either estradiol or an estrogen receptor (ER)alpha ligand has been shown to be both antiinflammatory and neuroprotective in a variety of neurological disease models, but whether neuroprotective effects could be observed in the absence of an antiinflammatory effect has remained unknown. Here, we have contrasted effects of treatment with an ERalpha vs. an ERbeta ligand in experimental autoimmune encephalomyelitis, the multiple sclerosis model with a known pathogenic role for both inflammation and neurodegeneration. Clinically, ERalpha ligand treatment abrogated disease at the onset and throughout the disease course. In contrast, ERbeta ligand treatment had no effect at disease onset but promoted recovery during the chronic phase of the disease. ERalpha ligand treatment was antiinflammatory in the systemic immune system, whereas ERbeta ligand treatment was not. Also, ERalpha ligand treatment reduced CNS inflammation, whereas ERbeta ligand treatment did not. Interestingly, treatment with either the ERalpha or the ERbeta ligand was neuroprotective, as evidenced by reduced demyelination and preservation of axon numbers in white matter, as well as decreased neuronal abnormalities in gray matter. Thus, by using the ERbeta selective ligand, we have dissociated the antiinflammatory effect from the neuroprotective effect of estrogen treatment and have shown that neuroprotective effects of estrogen treatment do not necessarily depend on antiinflammatory properties. Together, these findings suggest that ERbeta ligand treatment should be explored as a potential neuroprotective strategy in multiple sclerosis and other neurodegenerative diseases, particularly because estrogen-related toxicities such as breast and uterine cancer are mediated through ERalpha.