We reviewed the three reference toxicokinetic studies commonly used to suggestinnocuity of aluminum (Al)-based adjuvants. A single experimental study was carried out usingisotopic26Al (Flarend et al., 1997). This study ignored adjuvant cell capture. It was conductedover a short period of time (28 days) and used only two rabbits per adjuvant. At the endpoint,Al retention was 78% for aluminum phosphate and 94% for aluminum hydroxide, both resultsbeing incompatible with quick elimination of vaccine-derived Al in urines. Tissue distributionanalysis omitted three important retention sites: the injected muscle, the draining lymph nodeand bone. Two theoretical studies have evaluated the potential risk of vaccine Al in infants, byreference to the oral Minimal Risk Level (MRL) extrapolated from animal studies. Keith et al.,2002 used a too high MRL (2 mg/kg/d), an erroneous model of 100% immediate absorptionof vaccine Al, and did not consider renal and blood-brain barrier immaturity. Mitkus et al.(2011) only considered absorbed Al, with erroneous calculations of absorption duration. Theyignored particulate Al captured by immune cells, which play a role in systemic diffusion andthe neuro-inflammatory potential of the adjuvant. MRL they used was both inappropriate (oralAl vs injected adjuvant) and far too high (1 mg/kg/d) with regard to experimental studiesof Al-induced memory and behavioral changes. Both paucity and serious weaknesses of thesestudies strongly suggest that novel experimental studies of Al adjuvants toxicokinetics shouldbe performed on the long-term, including post-natal and adult exposures, to ensure innocuityand restore population confidence in Al-containing vaccines.