There is an urgent unmet need for reliable cardiac safety pharmacology assays to identify potential risks early in drug development and reduce time and cost to market. Human pluripotent (embryonic) stem cells (hESC) are a renewable, scalable and reproducible source of cardiomyocytes (CM) on which to base such test systems. Microarray analysis of modulations in gene expression during differentiation has shown that the major known cardiac genes are upregulated but that novel genes are also expressed. Normal cardiac development is recapitulated in vitro. Here we describe the field potential of hESC-CM, measured using commercially available multi electrode arrays. Systematic generation of dose response curves for cardiac and non-cardiac drugs show that hESC-CM accurately predicts reported drug effects on the human heart. These include blocking the human Ether-a-go-go Related Gene (hERG) ion channel, resulting in QT prolongation; this is associated with life-threatening arrhythmias, such as Torsade de Pointes (TdP). On this basis, we propose two directly applicable safety criteria for pre-clinical evaluation of new drugs in development: (1) prolongation of field potential duration (FPD) and (2) sodium peak reduction. This is the first study in which dose responses of such a wide range of compounds have been compared in hESC-CM and the outcome shown to predict clinical effects. We propose that assays based on hESC-CM could complement or potentially replace some of the preclinical tests currently used to select chemical compounds for development as new cardiac drugs and improve safety confidence once in clinical use.