1. In anaesthetized and artificially ventilated dogs, a gracilis muscle was vascularly isolated and perfused at a constant flow rate of 11.9 +/- 2.2 ml min-1 100 g-1 (mean +/- S.E.M., n = 16; equivalent to 170.2 +/- 21.3% of its resting free flow). 2. Stimulation (3 Hz) of the obturator nerve produced twitch contractions of the gracilis muscle, reduced venous pH from 7.366 +/- 0.027 to 7.250 +/- 0.031 (n = 5), increased oxygen consumption from 0.62 +/- 0.24 to 2.76 +/- 0.46 ml min-1 100 g-1 (n = 5) and increased adenosine release from -0.40 +/- 0.14 (net uptake) to 1.36 +/- 0.50 nmol min-1 100 g-1 (n = 8). 3. Infusion of lactic acid (4.2 mM) into the artery reduced venous pH to 7.281 +/- 0.026 (n = 5) and increased adenosine release to 0.96 +/- 0.40 nmol min-1 100 g-1 (n = 8), but did not significantly alter oxygen consumption (0.80 +/- 0.19 ml min-1 100 g-1; n = 5). Stimulation (3 Hz) in the presence of lactic acid infusion produced no further significant changes in venous pH or adenosine release, but increased oxygen consumption to 2.53 +/- 0.37 ml min-1 100 g-1 (n = 5). 4. Infusion of a range of lactic acid concentrations (> or = 1.83 mM) produced dose-dependent increases in adenosine release. The maximum lactic acid concentration tested (5.95 mM) reduced venous pH to 7.249 +/- 0.023 (n = 5) and increased adenosine release to 2.64 +/- 1.26 nmol min-1 100 g-1 (n = 6). 5. A strong correlation existed between the adenosine release and the venous pH (r = -0.92); points obtained during muscle stimulation and/or lactic acid infusion fell on a single correlation line. 6. The vasoactivity of adenosine administered by close-arterial injection was unaltered by infusion of either lactic acid (7.2 mM) or saline. 7. These results suggest that the release of adenosine from skeletal muscle can be induced by a decrease in pH (probably at an intracellular site), and that this mechanism may contribute to the release of adenosine during muscle contractions.