The tonA gene of Escherichia coli K12 was cloned into a multicopy plasmid, leading to substantial overproduction of the corresponding 78,000 Mr polypeptide in the outer membrane. The approximate size of the tonA gene and its direction of transcription were established by Tn1000 mutagenesis. A family of tonA deletions was constructed in vitro by Bal31 exonuclease digestion, followed by splicing of an "oligo stop" sequence to each 3' terminus in order to ensure prompt termination of translation of the truncated polypeptides in vivo. All these polypeptides proved to be extremely unstable in exponentially growing cultures but were relatively stable in maxicells. Under these conditions the truncated polypeptides, unlike wild-type TonA, fractionated with the Sarkosyl-soluble fraction of the cell envelope, indicating that these proteins are blocked in assembly as inner membrane (translocation) intermediates or as outer membrane (maturation) intermediates unable to form Sarkosyl-resistant complexes. We have also examined the kinetics of assembly of wild-type TonA into the outer membrane and the results indicate that, following cleavage of the N-terminal signal peptide, the protein passes through an apparently membrane-free intermediate form and only appears in the outer membrane after a delay of at least 50 seconds, following the completion of synthesis. From these results, we propose that the assembly of TonA involves translocation (with concomitant cleavage of the signal sequence) directly into the periplasm, followed by partitioning into the outer membrane. We further propose that the C terminus of TonA is essential for final maturation in the outer membrane in Sarkosyl-resistant form but that the C-terminal half of the molecule probably does not contain any topogenic sequences required for partitioning to the outer membrane.