The rate of exocellular levansucrase synthesis in an overproducing (sacUh) strain of Bacillus suhtilis was shown to be directly proportional to the amount of two different transient forms of this enzyme located within the membrane fraction of the cells. The apparent M , of the larger membrane form was 53000, and that of the smaller form 50000; the half-life time of each form was estimated in vivo to be 4-6 s and 32-42 s, respectively. Ethanol treatment of the cells lead to the accumulation of the 53 000-M, form which may represent 1.5% of total membrane proteins. This latter form, partially purified, was transformed in vitro into the 50000-M, form by the action of the Escherichia coli leader peptidase. These enzyme forms were quite different from the exocellular levansucrase since they showed a weak affinity for hydroxyapatite and needed complexed iron to display enzyme activity. Assuming the membrane forms were precursors of exocellular levansucrase, we propose a two-step mechanism for the secretion process of levansucrase. The number of exoprotein synthesis/secretion sites in a B. suhtilk cell is estimated to 2.5 x lo4. [4] the sacUh strain QBl12, when fully induced, produces only exocellular levansucrase (about 8% of total proteins) and, moreover, this enzyme became the unique protein secreted in the culture medium during the exponential phase of growth. Thus, this strain seems to contain optimal characteristics for a biochemical study of levansucrase secretion.The existence of a 53000-M, precursor form of levansucrase was demonstrated by the expression of the cloned structural gene of this enzyme in a minicell-producing strain of Escherichia coli; the nucleotide sequence of the signal peptide was determined [5, 61. The levansucrase precursor form synthesized in minicells in the presence of phenethyl alcohol was found in the outer membrane fraction [7]. However, the yield of the precursor form accumulated in such conditions (less than 0.02% of total protein) was too low for purification. Furthermore, characterization of the other components of a presumed B. subtilis export machinery is not possible using this approach. In order to prepare a large amount of the precursor of exocellular levansucrase, we investigated the conditions of its synthesis and accumulation in