The present study focuses on further extensions of the more general three-parameter Weibull distribution to describe the statistical scatter of fracture toughness values and to evaluate the characteristic toughness of structural steels using a statistical description of toughness data in comparison with the minimum of three equivalent (MOTE) method. Fracture toughness tests conducted on several types of structural steels, including an ultra high strength steel and pressure vessel steels, provide the experimental data upon which the Weibull statistical analyses are conducted. These analyses compare descriptions of fracture toughness values based on a standard three-parameter Weibull function with fixed values for parameters α and Kmin, and a general three-parameter Weibull distribution with unknown parameters (α,K0,Kmin) in connection with a goodness-of-fit method to assess how well the experimental data fits the assumed distribution. Further, the study also shows that use of a fixed percentile of the distribution describing the toughness data set provides more consistent values of characteristic toughness compared to the MOTE procedure. •Implications on the use of a 2P and 3P Weibull model for the scatter of toughness.•Use of Kolmogorov–Smirnov goodness-of-fit test to fracture toughness distributions.•Relatively weak dependence of T0 on the fitted Weibull distribution.•The MOTE method may provide unreliable evaluation of the characteristic toughness.•A 3P Weibull distribution gives a more consistent characteristic toughness value.