The heterodimeric pre-mRNA splicing factor, U2AF (U2 snRNP auxiliary factor), plays a critical role in 3′ splice site selection. Although the U2AF subunits associate in a tight complex, biochemical experiments designed to address the requirement for both subunits in splicing have yielded conflicting results. We have taken a genetic approach to assess the requirement for the Drosophila U2AF heterodimer in vivo. We developed a novel Escherichia coli copurification assay to map the domain on the Drosophila U2AF large subunit (dU2AF50) that interacts with the Drosophila small subunit (dU2AF38). A 28-amino-acid fragment on dU2AF50 that is both necessary and sufficient for interaction with dU2AF38 was identified. Using the copurification assay, we scanned this 28-amino-acid interaction domain for mutations that abrogate heterodimer formation. A collection of these dU2AF50 point mutants was then tested in vivo for genetic complementation of a recessive lethal dU2AF50 allele. A mutation that completely abolished interaction with dU2AF38 was incapable of complementation, whereas dU2AF50 mutations that did not effect heterodimer formation rescued the recessive lethal dU2AF50 allele. Analysis of heterodimer formation in embryo extracts derived from these interaction mutant lines revealed a perfect correlation between the efficiency of subunit association and the ability to complement the dU2AF50 recessive lethal allele. These data indicate that Drosophila U2AF heterodimer formation is essential for viability in vivo, consistent with a requirement for both subunits in splicing in vitro.