The recently determined three-dimensional structure of the pore-forming domain of colicin A has led to a hypothetical model for membrane insertion and channel formation. Certain features of this model have implications for understanding the mechanism of membrane insertion by other toxins and may have a broader relevance to protein transport in general.