A comparison of nonlinear vector network analyser (NVNA) measurements has been carried out involving four organisations (National Physical Laboratory, UK, University of Surrey, UK, Chalmers University of Technology, Sweden and Keysight Technologies, Denmark). Three nonlinear devices consisting of two amplifiers and a nonlinear verification device (NLVD) were measured by each of the organisations. Results are presented which show generally good agreement between the measurements and give some indication of the typical amount of variability to be expected in measurements of this type.