When adenosine diphosphate (ADP) is added to hirudinized platelet-rich plasma (PRP) in which the level of platelet cAMP has been pharmacologically elevated, there is an initial rapid fall in the level of cAMP brought about by inhibition of adenylate cyclase. This may be followed by a subsequent activation of adenylate cyclase that does not occur when citrated PRP is used in place of hirudinized PRP, and is more pronounced in the presence of added Mg2+. Here we provide evidence that a) the Mg2+-dependent activation of adenylate cyclase seen in hirudinized PRP is mediated by adenosine, b) the adenosine produced synergizes with forskolin and with DN9693 to raise the level of cAMP in platelets. but not with iloprost, c) Mg2+ does not influence directly the rate or extent of cAMP production and so is more likely to influence the rate of adenosine production, and d) activation of adenylate cyclase by adenosine can lead to inhibition of platelet aggregation. ARL 66096, a P2T purinoceptor antagonist which inhibits ADP-induced platelet aggregation, prevented inhibition of adenylate cyclase by ADP. Conversely, ARL 66096 did not appear to inhibit conversion of ADP to adenosine and subsequent activation of adenylate cyclase.