[en] A one-step optimization method based on a least squares fit of the linear quadratic model to quantitative tissue response data after fractionated irradiation is proposed. Suitable end-points that can be analysed by this method are growth delay, host survival and quantitative biochemical or clinical laboratory data. The functional dependence between the transformed dose and the measured response is approximated by a polynomial. The method allows for the estimation of the alpha/beta ratio and its confidence limits from all observed responses of the different fractionation schedules. Censored data can be included in the analysis. A method to test the appropriateness of the fit is presented. A computer simulation illustrates the method and its accuracy as examplified by the growth delay end point. A comparison with a fit of the linear quadratic model to interpolated isoeffect doses shows the advantages of the direct method. (orig./HP)