PurposeThe purpose of this study was to examine the boundary conditions of Kanter's (1977) tokenism theory as applied to the gender wage gap. The authors aimed to discover if there was a point where the relationship between the percentage of women in a job category and the gender wage gap changed, and, if so, where the threshold was located and what was the nature of the shift in relationship.Design/methodology/approachThe authors used the Andrews’ (1993) threshold effects technique. Using 22 separate years of publicly available Canadian wage data, they examined the relationship between the percentage of females in 40 unique occupational categories and the female-to-male earnings ratio (for a total of 880 observations).FindingsThe results showed the existence of a threshold point, and that early gains in percent female within an occupation, up to approximately 14% female in the occupation, associate with strong gains in the female-to-male wage ratio. However, beyond that point, further gains in percent female associate with smaller improvements in the female-to-male wage ratio.Practical implicationsThe findings are useful in understanding the dynamics of occupational group gender composition, potential theoretical reasons for the nuances in relationship, as well as opportunities that may facilitate more equitable outcomes.Originality/value The results show that, though improvements were made above and below the threshold point, enhancements in the wage gap are actually larger when there are less women in the job category (e.g. tokens).