Since the catastrophic Sumatra–Andaman tsunami took place in 2004, 16 other tsunamis have resulted in significant damage and 14 in casualties. We review the fundamental changes that have affected our command of tsunami issues as scientists, engineers and decision-makers, in the quest for improved wisdom in this respect. While several scientific paradigms have had to be altered or abandoned, new algorithms, e.g. the W seismic phase and real-time processing of fast-arriving seismic P waves, give us more powerful tools to estimate in real time the tsunamigenic character of an earthquake. We assign to each event a ‘wisdom index’ based on the warning issued (or not) during the event, and on the response of the population. While this approach is admittedly subjective, it clearly shows several robust trends: (i) we have made significant progress in our command of far-field warning, with only three casualties in the past 10 years; (ii) self-evacuation by educated populations in the near field is a key element of successful tsunami mitigation; (iii) there remains a significant cacophony between the scientific community and decision-makers in industry and government as documented during the 2010 Maule and 2011 Tohoku events; and (iv) the so-called ‘tsunami earthquakes’ generating larger tsunamis than expected from the size of their seismic source persist as a fundamental challenge, despite scientific progress towards characterizing these events in real time.