1. Spatiotemporal aspects of subcellular Ca2+ signalling were studied in cultured adult guinea-pig atrial myocytes. A mixture of the Ca2+ indicators fluo-3 and Fura Red in combination with laser-scanning confocal microscopy was used for [Ca2+]i measurements while membrane currents were recorded simultaneously. 2. In citrate-loaded atrial myocytes not every Ca2+ current (ICa) could trigger Ca2+ release from the sarcoplasmic reticulum (SR). Two types of Ca2+ signals could be observed: Ca2+ transients resulting from (i) Ca2+ influx alone and (ii) additional Ca2+ release. 3. Ca2+ release elicited by voltage steps of 100-150 ms duration was either apparently homogeneous or propagated as Ca2+ waves through the entire cell. With brief ICa (50-75 ms), Ca2+ waves with limited subcellular propagation were observed frequently. These waves always originated from either end of the myocyte. 4. The time course of changes in Na(+)-Ca2+ exchange current (INaCa) depended on the subcellular properties of the underlying Ca2+ transient and on the particular cell geometry. Apparently homogeneous Ca2+ release was accompanied by an inward change of INaCa the onset phase of which was fused with ICa. Changes in INaCa caused by a Ca2+ wave propagating through the entire cell showed a W shape, which could be attributed to differences of the fractional surface-to-volume ratio in different cell segments during propagation of the Ca2+ wavefront. Those waves with limited spreading only activated a small component of INaCa. 5. The different subcellular patterns of Ca2+ release signals can be explained by spatial inhomogeneities in the positive feedback of the SR. This depends on the local SR Ca2+ loading state under the control of the local Ca2+ influx during activation of ICa. Due to the higher surface-to-volume ratio at the two ends of the myocyte, SR loading and therefore the positive feedback in Ca(2+)-induced Ca2+ release may be higher at the ends, locations where Ca2+ waves are preferentially triggered. 6. We conclude that the individual cell geometry may be an important determinant of subcellular Ca2+ signalling not only in cardiac muscle cells but presumably also in other types of cells that depend on Ca2+ signalling. In addition, the cell geometry in combination with varying subcellular Ca2+ release patterns can greatly affect the time course of Ca(2+)-activated membrane currents.