Multihypothesis pictures are related to B pictures where two motion-compensated signals from previous and future reference frames are linearly combined for prediction. Multihypothesis pictures also linearly combine two motion-compensated signals but only from temporally previous reference frames. Consequently, multihypothesis pictures have the advantage that there is no extra coding delay like for B pictures. In addition, multihypothesis pictures are also used for reference to predict future multihypothesis pictures. We integrate multihypothesis prediction into the emerging H.26L standard by introducing one additional multihypothesis macroblock mode for low delay video coding which is similar to the bi-directional macroblock mode for B pictures. The current implementation achieves bit-rate savings up to 13% for the investigated test-sequences when compared to the H.26L test model.