Within the SAM project, special software has been developed for scoring the results of a recognition test. The program is called ``SAM_SCORE '', and is designed for small vocabulary systems (see Appendix E). It is able to determine confidence limits for the performance measures and do a McNemar test on the difference in performance of two recognisers . The program can both be used in batch mode and in interactive mode with a graphical user interface. It can produce output in a DBMS format, which can be used for subsequent analysis software (see below and Appendix H on DBMSs).