Speech recogniser assessment, an area of speech recognition research which is generating a great deal of interest. Currently available assessment techniques tend to produce inaccurate or unreliable results, this has been seen to be due to essential limitations in the methodologies adopted. This paper describes an assessment methodology designed to overcome many of the limitations of the current techniques. It has a mathematical basis and uses a statistically derived data base to develop standard tests for various environments of interest. It is hoped that by the use of these techniques experimenters will be able to perform laboratory based recogniser tests with confidence. It is expected that for many applications these laboratory tests will not replace field trials, but will provide guidence towards suitable recognisers.