Xenon isotopic data were acquired by high resolution step pyrolysis and combined step pyrolysis/combustion of aliquots of size separated nanodiamonds. 129Xe excess (129Xe*) from in situ decay of 129I is preferentially associated with the larger grain size separates. This observation rules out trapping by recoil from surrounding material. The releases of Xe-P3 and 129Xe occur in the same low temperature pyrolysis steps and exhibit similar distributions among the size separates. These observations imply a common site for the components and, in consequence, suggest a common incorporation event. Whether one component or two, our observations require that 129Xe* and Xe-P3 were incorporated into a subpopulation of nanodiamonds before nanodiamonds were mixed and incorporated into parent bodies. Their susceptibilities to loss during heating in the laboratory are similar, but the ratio of 129Xe* to Xe-P3 varies among nanodiamond separates from different meteorites (literature data). We conclude that the 129Xe* we observe today was present as 129I during parent body processing. Furthermore, the range of 129Xe*/ 132XeP3 ratios across all the separates requires that even nanodiamonds from CI chondrites were at least 5-10× more rich in Xe-P3 during 129I decay than they are today. We present a simple model involving one degassing event per parent body between incorporation of nanodiamonds and final decay of 129I. The observed variations among parent bodies require degassing events separated by several 129I half lives (∼50Ma), consistent with low-temperature processing on parent bodies but longer than expected for nebular processing. In this model, nanodiamonds from ALHA77307 degassed at an unusually early stage, suggesting they alone may retain the signature of processing in the nebula in their P3 and 129Xe* abundances. The isotopic signature associated with Xe-P6 is also found only in the larger size separates. Concentration of Xe-HL increases with increasing grain size, but its relative abundance with respect to Xe-P3 and P6 is higher in smaller grain-size fractions. We argue that Xe-P6 is best seen as a variant of Xe-HL, and that they are both mixtures of a "normal" component akin to solar xenon and a slightly variable exotic component. We show that both current models of Xe-H formation can account for the observed variability, and propose a scenario according to which Xe-HL and P6 were implanted into separate diamond populations before incorporation of Xe-P3 and 129I. Copyright © 2005 Elsevier Ltd.