sWe present a semi-numerical method to compute one-loop corrections. We apply the method to the calculation of virtual six-gluon amplitudes.