To determine whether pain or discomfort could be provoked by adenosine in skeletal muscle and, if so, whether it was dependent on the vasodilatation produced by adenosine, eight male volunteers were given intra-arterial bolus injections of adenosine and glyceryl trinitrate into the forearm. Local pain was assessed on a scale rate, forearm blood flow was measured by venous occlusion plethysmography, and blood was sampled simultaneously from the deep vein of the same arm. Five different doses of adenosine, ranging between the maximum tolerable and the lowest producing pain or discomfort, were given intra-arterially in random order and repeated in reverse order. Glyceryl trinitrate was given intra-arterially in increasing doses from 1 to 20 micrograms. Pain or discomfort began 12(1)(SEM) s after administration reached its maximum after 17(1) s, and disappeared after 40(2) s. Pain or discomfort appeared 8(1) s (p less than 0.001) after the first recorded increase in forearm blood flow, whereas maximum pain or discomfort preceded maximal forearm blood flow by 5(1) s (p less than 0.001). The flow remained increased after the disappearance of pain or discomfort. The effects of adenosine on pain or discomfort and vasodilatation were dose dependent. Glyceryl trinitrate provoked a similar increase in flow to that with adenosine without producing pain or discomfort. Arterial occlusion for 5 min at rest or forearm exercise with arterial occlusion increased forearm blood flow to the same extent as the maximum dose of adenosine. In addition, ischaemic work slightly increased the plasma concentration of adenosine. The pain or discomfort after ischaemic work was not considered different from the adenosine provoked pain or discomfort by four of the subjects. It is concluded that the symptoms did not appear to be dependent on vasodilatation and therefore adenosine may contribute to ischaemic pain or discomfort.