The hypothesis that platelet ADP is responsible for collagen-induced aggregation has been re-examined. It was found that the concentration of ADP obtaining in human PRP at the onset of aggregation was not sufficient to account for that aggregation. Furthermore, the time-course of collagen-induced release in human PRP was the same as that in sheep PRP where ADP does not cause release. These findings are not consistent with claims that ADP alone perpetuates a collagen-initiated release-aggregation-release sequence. The effects of high doses of collagen, which released 4-5 microM ADP, were not inhibited by 500 microM adenosine, a concentration that greatly reduced the effect of 300 microM ADP. Collagen caused aggregation in ADP-refractory PRP and in platelet suspensions unresponsive to 1 mM ADP. Thus human platelets can aggregate in response to collagen under circumstances in which they cannot respond to ADP. Apyrase inhibited aggregation and ATP release in platelet suspensions but not in human PRP. Evidence is presented that the means currently used to examine the role of ADP in aggregation require investigation.