Mammalian splicing factor SF1 consists of a single polypeptide of 75 kDa and is required for the formation of the first ATP-dependent spliceosomal complex. Three cDNAs encoding variant forms of SF1 have been isolated and four highly related cDNAs have been found in current databases. Comparison of the cDNA sequences suggests that different SF1 mRNAs are generated by alternative splicing of a common pre-mRNA. In agreement with this idea, at least three mRNAs that are differentially expressed in different cell types have been detected by northern blot analysis. All SF1 cDNAs identified encode proteins with a common N-terminal half that contains two structural motifs implicated in RNA binding (an hnRNP K homology [KH] domain and a zinc knuckle), but the proteins differ in the length of a proline-rich region and have distinct C-termini. Three SF1 isoforms expressed in insect cells via baculovirus transfer vectors show comparable activities in the assembly of a pre-splicing complex. Consistent with the presence of a KH domain and a zinc knuckle, we show that SF1 binds directly to RNA. This interaction appears to be largely sequence-independent with a preference for guanosine- and uridine-rich sequences. The KH domain of SF1 is embedded in a 160-amino acid sequence that is shared with human Sam68, a target of Src during mitosis, as well as Caenorhabditis elegans GLD-1 and mouse Qkl, both of which play roles during cellular differentiation. The presence of this shared region in SF1 suggests functions in addition to its role in pre-spliceosome assembly.