Techniques are provided for use by implantable medical devices for controlling ventricular pacing. In one example, optimal atrio-ventricular and interventricular pacing delay values are determined for pacing the heart of the patient based, in part, on a measured inter-atrial conduction delay. Atrio-ventricular conduction delays are then measured within the patient. The atrio-ventricular pacing delays are compared with the measured atrio-ventricular conduction delays. If the atrio-ventricular pacing delays are less than the measured atrio-ventricular conduction delays, biventricular pacing is delivered using the atrio-ventricular pacing delay and the interventricular pacing delay. However, if the atrio-ventricular pacing delays are not less than the corresponding atrio-ventricular conduction delays, as can occur if the inter-atrial conduction delay is large, then alternative pacing regimes are selectively enabled, such as monoventricular pacing in the chamber having the longer conduction delay value, biventricular pacing with negative hysteresis, or biventricular pacing with pacing delays reduced using predetermined offset values.