This study re-examines the relationship between women on corporate boards (WOCB) and environmental performance (EP) for a sample of firms from the Fortune 1000 2020 list over the period 2004-2020 (17 years). To account for the nonlinearity between WOCB and EP and the endogeneity issues associated to this relationship, we consider semiparametric specifications, and specifically generalized additive models (GAM; Wood, 2006). Specifically, we use Marra and Radice's (2011) two-stage generalized additive model (2SGAM). We find that the effects of WOCB on EP depend significantly on their numerical representation, thereby suggesting a nonlinear relationship between WOCB and EP. This is consistent with Kanter's (1977) regarding token and critical mass theories. Our findings challenge the existing theoretical empirical studies on this matter.