Abstract The described procedure starts with the mathematical fitting of a survival non-linear equation to experimental points by the least squares method, including a remark about the statistical weight attached to each point. Then it permits a test of the goodness of the fit and estimations of its absolute and relative precisions. It gives also an equation for the confidence zone limits taking into account the survival heteroscedasticity. The use of a non-parametric simultaneous test procedure makes possible comparisons of the precisions achieved with several models fitted to the same numerous experimental curves. Finally, it suggests an estimation of the coefficient of variation of each of the parameters fitted to several curves. As an example, this procedure is applied to the analysis of four models fitted to 208 survival curves of chlorella cells exposed to gamma or particulate radiations.