A model based high-resolution QRS fiducial point correction algorithm, which is suitable for sparsely sampled electrocardiogram (ECG) recordings, is presented. The presented method can be divided into three steps. First, the initial QRS fiducial points are estimated by using ordinary interpolation methods. Then, the data points of each QRS are extracted and centered in time and the shape of the QRS complex is estimated by nonlinearly fitting a double exponential function to the extracted data points. Finally, the estimated model and its derivative are linearly fitted to the data points of each QRS complex separately and new fiducial point estimates are obtained. The proposed method is tested with simulations and real ECG data. As a result, it is observed that the proposed method is also suitable for asymmetric QRS complexes unlike, e.g., the commonly used cubic spline interpolation method.