Following previous studies on the associations between aluminium (Al) water concentrations and relatively high risks of a measure of mental impairment, the role of silica (SiO2) as a possible confounding variable is investigated. The bivariate associations show that the lowest odds, representing an approximation of relative risks, are observed at “mild” SiO2 concentrations. Multivariate analyses indicate that, if interaction terms are included, the odds comparing higher and lower SiO2 concentrations are different than if an interaction term is not included. There are also indications that at relatively high Al levels, high SiO2 concentrations lower the odds, consistent with previous suggestions by Birchall. The implication is that when investigating the role of Al in the etiology of various measures of mental impairment, it is important to consider the effects of other water constituents, and also the different chemical forms of Al present in treated water. The desirability of additional work is stressed, preferably using more definitive outcomes than indications of mental impairment.