A new numerical method is presented for calculating the particle‐size distribution function of dilute particle systems from Small Angle Scattering data. The size distribution function, in order to be always positive, is approximated by the square of a combination of cubicB‐splines. The coefficients of the combination are calculated in reciprocal space using the weighted least squares principle, which leads to a variational problem, solved by a Newton‐Gauss‐Marquardt iterative technique. A strategy is presented to enhance resolution in correspondence to the peaks of the distribution. Some results obtained on simulated and experimental points are presented, showing the accuracy of the method.