Clinopyroxene inclusions in diamond contain elevated potassium contents and can potentially be dated by 40Ar/39Ar techniques. Previous 40Ar/39Ar studies of clinopyroxene inclusions contained in cleaved diamonds have suggested that argon, produced from the decay of potassium prior to eruption of the host kimberlite magma, diffuses to the diamond/clinopyroxene interface under mantle conditions. After intrusion and cooling below the closure temperature for argon diffusion, radiogenic argon is retained by the clinopyroxene inclusions. This behaviour complicates efforts to date diamond crystallisation events; however, extraction of inclusions from their host diamond should induce loss of all interface argon, thus raising the possibility of determining kimberlite emplacement ages. This possibility has important implications for constraining the source localities of detrital diamond deposits worldwide, with concomitant benefits to diamond exploration. To investigate this premise, 40Ar/39Ar laser probe results are presented for single clinopyroxene inclusions extracted from a total of fifteen gem-quality diamonds from the Mbuji-Mayi kimberlite in the Democratic Republic of Congo, and the Jwaneng and Orapa kimberlites in Botswana. Initial fusion analyses of clinopyroxene inclusions from Mbuji-Mayi diamonds yielded ages older than the time of host kimberlite intrusion, indicating partial retention of extraneous argon by the clinopyroxene inclusions themselves. Step-heating analyses of clinopyroxene inclusions from Orapa and Jwaneng diamonds produced older apparent ages from lower temperature steps and the 'rim' fragment of one Orapa inclusion. High temperature (fusion) analyses yielded younger apparent ages, commonly approaching the times of host kimberlite eruption. Total-gas integrated 40Ar/39Ar ages are mostly intermediate between the times of inferred diamond crystallisation and kimberlite eruption. Ca/K ratios for each sample are uniform across step-heating increments, indicating that age variations are not due to compositional, mineralogical or alteration effects. The favoured explanation for these results is partial retention of extraneous argon in primary and/or secondary fluid inclusions. This component is then preferentially outgassed in lower temperature heating steps, yielding older apparent ages.The partial retention of extraneous argon by clinopyroxene inclusions clearly restricts efforts to determine source ages for detrital diamond deposits. Results from individual samples must necessarily be interpreted as maximum source emplacement ages. Nonetheless, step-heating analyses of several clinopyroxene inclusions from a detrital diamond deposit may provide reasonable constraints on the ages of source kimberlites/lamproites; however minor age populations as well as those closely spaced in time, may be difficult to resolve. It is argued that the majority of older 40Ar/39Ar ages can be explained in terms of the partial retention of inherited argon, produced between the times of diamond crystallisation and kimberlite eruption. Although the presence of excess argon in some clinopyroxene inclusions cannot be excluded, available evidence (e.g. no excess argon in Premier eclogitic inclusions or potassium-poor inclusions) suggests that this is not a factor for most samples. Three possible mechanistic models are forwarded to account for the uptake of inherited (± excess) argon in fluid inclusions. The first envisages negligible interface porosity and diffusion of extraneous argon exclusively to primary fluid inclusions, which subsequently partially decrepitated during eruption, causing accumulation of argon at the diamond/clinopyroxene interface. The second model permits diffusive loss of extraneous argon to both the interface region and primary fluid inclusions. The third involves diffusion of extraneous argon to the interface region, with later entrapment of some interface argon in secondary fluid inclusions, produced by fracture/annealing processes active during eruption. The first model can account for all 40Ar/39Ar results, whereas the latter two mechanisms require the presence of an excess argon component to explain older integrated ages (up to 2.9 Ga) from two Jwaneng samples. Excess argon contamination would compromise efforts to determine diamond genesis ages using the 40Ar/39Ar dating technique. However, if the first model is valid, then the older 40Ar/39Ar integrated ages support previous Re-Os age results for the crystallisation of Jwaneng diamonds.