An important step in the biosynthesis of many proteins is their partial or complete translocation across the plasma membrane in prokaryotes or the endoplasmic reticulum membrane in eukaryotes. In bacteria, secretory proteins are generally translocated after completion of their synthesis by the interaction of the cytoplasmic ATPase SecA and a protein-conducting channel formed by the SecY complex. How SecA moves substrates through the SecY channel is unclear. However, a recent structure of a SecA-SecY complex raises the possibility that the polypeptide chain is moved by a two-helix finger domain of SecA that is inserted into the cytoplasmic opening of the SecY channel. Here we have used disulphide-bridge crosslinking to show that the loop at the tip of the two-helix finger of Escherichia coli SecA interacts with a polypeptide chain right at the entrance into the SecY pore. Mutagenesis demonstrates that a tyrosine in the loop is particularly important for translocation, but can be replaced by some other bulky, hydrophobic residues. We propose that the two-helix finger of SecA moves a polypeptide chain into the SecY channel with the tyrosine providing the major contact with the substrate, a mechanism analogous to that suggested for hexameric, protein-translocating ATPases.