Recent studies of India-Eurasia convergence suggest that the entire convergence in the Himalayan wedge is almost exclusively accommodated along its basal detachment fault (Main Himalayan thrust, MHT) and its near-surface equivalent (Main Frontal thrust, MFT). Using direct dating of fault-zone gouge and strath terrace deposits, we conclude the following. (1) The present mountain front in the Darjiling sub-Himalaya was emplaced by ca. 40 ka. (2) Out-of-sequence deformation on surface-breaking faults north of the MFT in the Darjiling sub-Himalaya began ca. 20 ka and has probably continued since. (3) The Tista River responded to the ca. 20 ka deformation by migrating 150 m eastward (average rate ∼13 mm yr−1) and by incising 48 m vertically (average rate ∼4.4 mm yr−1), creating unpaired, disjointed strath terraces between 11.3 ± 1.3 ka and 1.4 ± 0.3 ka. Out-of-sequence, surface-breaking faults in the Himalaya indicate partial accommodation of active convergence within the Himalayan wedge. Using the results from the Bhuj earthquake of 2001, we suggest that active deformation along the out-of-sequence faults is a potential seismic hazard in the Himalaya, and Himalayan seismic hazard models must account for this. We also propose a conceptual model for active deformation in the Himalaya.