Studies on glial cultures have demonstrated that fetal bovine serum contains a factor that induces bipotential glial precursors known as oligodendrocyte-type 2 astrocyte (O-2A) progenitors to become type 2 astroglia rather than oligodendroglia. The goal of this research project was to characterize and purify this factor, which we refer to as the astroglia-inducing molecule (AIM). Using cultures enriched in O-2A progenitors, we determined that AIM is present in human and bovine sera and that fetal bovine serum qualified as the best serum for purifying AIM. AIM is heat and trypsin labile and may be a plasma glycoprotein. A 240-fold enriched AIM preparation was produced by applying an ammonium sulfate precipitate of fetal bovine serum to heparin and then lentil lectin-agarose, followed by gel filtration chromatography. In crude preparations, AIM activity migrated at 50 kDa by gel filtration. With enrichment, activity was seen at several molecular masses, all of which were approximate multiples of 50 kDa. Treatment with 6 M guanidine hydrochloride generated an AIM with a molecular mass between 12 and 18 kDa, a result suggesting that AIM aggregates. On a preparative isoelectric focusing gel, AIM activity most frequently migrated between pH values of 3 and 4; however, proteins with isoelectric points of greater than 9 or at 6 also had activity in several experiments. These data suggest that either multiple AIMs exist or that a single AIM exists that associates with other proteins. Immunofluorescence for ganglioside GD3 and glial fibrillary acidic protein confirmed that AIM preparations induce type 2 astroglia from O-2A progenitors and suggests that AIM has little effect on type 1 astroglia. Because none of the known growth factors that have been tested to date mimics its effects. AIM may be a novel differentiation factor.