A morphology discrimination scheme extracts shape characteristics from cardiac signals and identifies an associated cardiac condition based on the shape characteristics. For example, internal data structures may be updated to match the shape characteristics of a known condition (e.g., a patient's normal sinus rhythm). Similarly acquired shape characteristics obtained in conjunction with a later event (e.g., QRS complexes acquired during a tachycardia episode) may be compared with the previously stored shape characteristics to characterize the later event. In some aspects the shape characteristics relate to inflection points of cardiac signals.