A novel methodology for extrapolation ofthe performance of ITER class machines is introduced. This procedureutilizes the ITER ELMy H mode database by means of a similarityapproach where every discharge is extrapolated to a machine satisfying theITER performance requirements through the use of the tokamak systemcode developed during the course of the ITER design evolution. Thisapproach attempts, amongst other things, to overcome the difficultyassociated with the simultaneous choice of non-dimensional parameterswhich, in particular when close to their respective limits, may havesome significant mutual interactions affecting energy confinementtime. The methodology is also applied to the ITER Physics Basis scalingsto develop a non-statistical approach where fusionpower is extrapolated at constant beta and confinement is assumedto follow a gyroBohm form. In both cases, out of more than a thousanddischarges in the ELMy H mode database, less than half turn out toextrapolate to a Q = 10 machine whose major radius is smaller than8 m. However, a significant number of discharges do extrapolate to a Q = 10 machinewith R<6.2 m. This strengthens confidence in the present choice ofITER parameters. In addition, from this analysis, it has beenpossible to identify a set of high performance `ITER relevant' dischargesfrom a number of machines, which could be used as starting points forinvestigation of further improvements in confinement.