Experimental design and Response Surface Methodology (RSM) were used to optimize the production of a dagger N-123-GBD-CD2, an alpha-(1 -> aEuro parts per thousand 2) branching sucrase previously reported as mainly produced in inclusion bodies. The a dagger N-123-GBD-CD2 encoding gene was cloned into two expression vectors in fusion with 6xHis tag or Strep tag II encoding sequences at 5' and 3' ends of the gene and expressed in five Escherichia coli strains. Three host-vector combinations were first selected on the basis of the amount of soluble enzyme produced. RSM with Box-Behnken design was used to optimize the expression conditions in an auto-inducible medium. Five factors were considered, i.e. culture duration, temperature and the concentrations of glycerol, lactose inducer and glucose repressor. The design consisted of three blocks of 45 assays performed in deep well microplates. The regression models were built and fitted well to the experimental data (R (2) coefficient > 94 %). The best response (production level of soluble enzyme) was obtained with E. coli BL21 Star DE3 cells transformed with the pET-55 vector. Using the predicted optimal conditions, 5,740 U L-1 (of culture) of soluble enzyme was produced in microtiter plates and more than 12,000 U L-1 (of culture) in Erlenmeyer flask, which represents a 165-fold increase compared to the production levels previously reported.