A method and apparatus for predicting acute response to cardiac resynchronization therapy is disclosed. The method can comprise measuring a first interval during an intrinsic systolic cycle and measuring a second interval during a stimulated systolic cycle. The acute response can be predicted by comparing the percent change in duration between the first interval and the second interval against a pre-determined threshold value. The first and second time intervals can be measured using, for example, a surface ECG or, alternatively, an intracardiac electrogram. In one embodiment, the first interval can be the duration of an intrinsic QRS complex measured during a non-stimulated systolic cycle. Similarly, the second interval can be the duration of a stimulated QRS complex measured during a stimulated systolic cycle.