Soluble silica reduces aluminum availability and ameliorates toxicity in several biological systems. It has therefore been suggested that these two species strongly interact in solution. However, there is only weak affinity between monomeric silicic acid and aluminum with reported log K-eff of between 4 and 6 at pH 7.2. We now show the existence of a soluble low molecular weight form of silica that is nonmonomeric but has an affinity for aluminum at least 1 000 000 times greater than the monomeric form (log K-eff = 11.70 +/- 0.30 at pH 7.2). This was established by competition, for binding of aluminum, between different preparations of soluble silica and the powerful M3+ chelator, 1,2-dimethyl-3-hydroxy-4-pyridinone (DMHP). At pH 7.2, this nonmonomeric silica quantitatively displaced one equiv of DMHP for every 34.42 +/- 0.77 equiv of total silica, suggesting that the soluble silica that so avidly binds aluminum is an oligomer containing fewer than 35 silicon atoms. The presence of oligomeric or monomeric silica in solution depends on how the solution is prepared. The oligomer is formed when alkali stock solutions of silica (42 mM in this work) are pH neutralized prior to dilution, to physiological and natural levels of soluble silica (<2.5 mM). The diluted oligomeric silica is at least transiently stable (<1 day or >17 days; absence or presence of aluminum) and would compete effectively with endogenous physiological chelators for aluminum, such as citrate (log K-eff aluminum citrate = 7.15 at pH 7.2). Oligomeric silica is probably responsible for the many experimental observations on the amelioration of aluminum availability and toxicity, but whether this silica species occurs naturally in the environment or is formed in vivo, remains to be established.