1. We sought to evaluate the contribution of the sarcoplasmic reticulum (SR) Ca2+ pump (vs. diffusion) to the kinetics of [Ca24]i decline during Ca2+ sparks, which are due to spontaneous local SR Ca2+ release, in isolated rat ventricular myocytes measured using fluo-3 and laser scanning confocal microscopy. 2. Resting Ca2+ sparks were compared before (control) and after the SR Ca2(+)-ATPase was either completely blocked by 5 microM thapsigargin (TG) or stimulated by isoprenaline. Na(+)-Ca2+ exchange was blocked using Na(+)-free, Ca(2+)-free solution (0 Na+, O Ca2+) and conditions were arranged so that the SR Ca2+ content was the same under all conditions when Ca2+ sparks were measured. 3. The control Ca2+ spark amplitude (281 +/- 13 nM) was not changed by TG (270 +/- 21 nM) or isoprenaline (302 +/- 10 nM). However, the time constant of [Ca2+]i decline was significantly slower in the presence of TG (29.3 +/- 4.3 ms) compared with control (21.6 +/- 1.5 ms) and faster with isoprenaline (14.5 +/- 0.9 ms), but in all cases was much faster than the global [Ca2+]i decline during a control twitch (177 +/- 10 ms). 4. The spatial spread of Ca2+ during the Ca2+ spark was also influenced by the SR Ca2+ pump. The apparent 'space constant' of the Ca2+ sparks was longest when the SR Ca2+ pump was blocked, intermediate in control and shortest with isoprenaline. 5. We conclude that while Ca2+ diffusion from the source of Ca2+ release is the dominant process in local [Ca2+]i decline during the Ca2+ spark, Ca2+ transport by the SR contributes significantly to both the kinetics and spatial distribution of [Ca2+]i during the Ca2+ spark.