We examine data from over 6.6 million games of tournament chess between players rated by the international chess authority, FIDE. Previous research has focussed on the low representation of women in chess. We replicate and extend previous analysis (Chabris and Glickman, 2006) on an international level. We find no support for differential variability, differential drop-out between male and female players, or social context (in the form of proportion of female players at a national level) as drivers of drivers of male-female differences. Further, we examine games between mixed and same gender pairs for evidence of a `stereotype threat' effect. Contrary to previous reports, we find no evidence of stereotype threat. Though this analysis contradicts one specific mechanism whereby gender stereotype may influence players, the persistent differences between male and female players suggests that systematic factors do exist and remain to be uncovered.