The aim of the current study was to identify the Rugby game- related statistics that discriminated between winning and losing teams in IRB and S12 close games. Archival data reported to game-related statistics from 120 IRB games and 204 Super Twelve games played between 2003 and 2006. Afterwards, a cluster analysis was conducted to establish, according to game final score differences, three different match groups. Only the close games group was selected for further analysis (IRB n = 64 under 15 points difference and Super Twelve n = 95 under 11 points difference). An analysis to the structure coefficients (SC) obtained through a discriminant analysis allowed to identify the most powerful game-related statistics in discriminating between winning and losing teams. The discriminant functions were statistically significant for Super Twelve games (Chi-square = 33.8, p < 0.01), but not for IRB games (Chi- square = 9.4, p = n.s.). In the first case, winners and losers were discriminated by possessions kicked (SC = 0.48), tackles made (SC = 0.45), rucks and pass (SC = -0.40), passes completed (SC = 0. 39), mauls won (SC = -0.36), turnovers won (SC = -0.33), kicks to touch (SC = 0.32) and errors made (SC = -0.32). The minus sign denotes higher values in losing teams. Rugby game-related statistics were able to discriminate between winners and losers in Super Twelve close games and suggest that a kicking based game supported by an effective defensive structure is more likely to win matches than a possession based one.