5-Hydroxytryptamine (5-HT) increases human heart rate and atrial contractile force and hastens atrial relaxation through 5-HT4 receptors. Moreover, 5-HT may be arrhythmogenic and give rise to atrial fibrillation. It is not clear which splice variant(s) of the 5-HT4 receptor is expressed and mediates these effects of 5-HT in the human heart. Previous studies have indicated different pharmacological properties of 5-HT4 receptors in human heart and mouse colliculi neurones, possibly due to expression of different splice variants. We therefore cloned the human 5-HT4(b) receptor and compared its pharmacological properties with those of the cloned human 5-HT4(a) receptor and searched for the corresponding mRNA in human tissues. The primary structures of the two human 5-HT4 receptor splice variants are identical except for divergent C-terminal tails of 28 and 29 amino acids in the 5-HT4(a) and 5-HT4(b) receptors, respectively. Reverse transcription-polymerase chain reaction (RT-PCR) analysis showed that both variants were coexpressed in various tissues, including cardiac atrium and ventricle. Additional bands suggested the presence of more than two human 5-HT4 receptor splice variants. With cloned receptors stably expressed in HEK293 cells or transiently expressed in COS-7 cells, [3H]GR 113808 consistently showed slightly higher binding affinity to h5-HT4(b) than to h5-HT4(a) receptor (pKd 0.1-0.2 log units higher). Competition of agonists, partial agonists and antagonists for [3H]GR113808 binding revealed no significant differences between the two receptors. For 5-HT4 receptor agonists and antagonists, their potencies in stimulating or inhibiting, respectively, 5-HT-stimulated adenylyl cyclase activity correlated well with their binding affinities. Tropisetron and SB207710 showed partial agonist activity at high receptor expression levels for both isoforms. Cisapride and renzapride were both partial agonists at moderate receptor levels and full agonists at high receptor levels. Cisapride was more potent than renzapride while the converse was the case in human atrium, for which cisapride had lower affinity and agonist potency than at the recombinant receptors. The binding affinities and agonist potencies of ligands for both 5-HT4(a) and 5-HT4(b) receptors correlated with the corresponding affinities and potencies in human atrium. The agonist potency of SB207710 was around 10 times lower than its binding and blocking affinity for both splice variants, suggesting that activation of adenylyl cyclase and blockade of 5-HT responses are mediated through different conformational states. The similar pharmacological properties of the two human 5-HT4 receptor splice variants together with their expression in human atrium would be consistent with mediation of the cardiostimulant effects of 5-HT through both variants. However, the effects of cisapride appear either mediated through non-a and non-b splice variants of the 5-HT4 receptor or 5-HT4(a) and 5-HT4(b) receptor expression in human atrial cells alters somewhat their pharmacological profile through still unknown mechanisms.