Temporal complexity of early ventricular fibrillation (VF) is re-assessed through measuring the correlation dimension D 2 , entropy K 2 and high-dimensional component σ from electrocardiogram (ECG) recordings using the Gaussian kernel algorithm. Seven representative ECG traces of induced VF among 53 pig subjects are selected for analysis. VF is found to have 80-90% low-dimensional deterministic dynamics with D 2 varying around 5 and 10-20% high-dimensional extents.