Records of the reproductive performance of boars on a 250 sow commercial pig herd were collated for four years. A new computer program was constructed and used for data on the daily liveweight gain and carcase grading of the whole herd and of individual sows and boars for selected periods. Two separate series of records, each for 11 boars, are presented and analysed statistically to determine significant differences between the boars in each series for farrowing rates to first service, total numbers born and liveweight gain of progeny. The overall farrowing rate to first service was 85.6 per cent for boars in series 1 and 82.1 per cent for those in series 2. One boar was found to have a farrowing rate to first service consistently below 75 per cent and was culled. The decision to cull was taken although its performance was only significantly different in series 1. No statistical differences between boars were found for numbers born. Statistical differences between boars were found for liveweight gain of progeny and the use of Duncan's multiple range test on ranked means in the analysis of such data is discussed. The necessity for the use of statistical analysis rather than simple ranking in the presentation of boar data is discussed and guidelines are given.