How do you prepare for tsunamis that come only every 1,000 years or so?
This is the issue Swiss geoscientists are wrestling with as they study the country's big lakes.
Some of these water bodies around the Alps have been known to experience huge waves that were driven by sub-surface landslides, which were themselves triggered by earthquakes.
The researchers' work indicates such hazards still exist but the likelihood of future events is very small.
"These incidents happen much less frequently than flooding or avalanches, which makes tsunamis a very hard-to-grasp hazard that is simply not in the minds of the population. But as geoscientists we have a duty to look at the problem and to let people know the risks," explained Prof Flavio Anselmetti from the University of Bern.
He was speaking here in Vienna at the European Geosciences Union General Assembly.
The professor recalled the estimated Magnitude 5.9 quake in the Alps region in AD 1601 that prompted sediment on sub-surface slopes in Lake Lucerne to slide down on to the lake basin floor.
As the muds descended, they alternately made the water column above slump and then rise, setting off a sequence of waves that inundated coastal communities.
The chronicles from the time talk of mountains of water appearing in the middle of the lake, and of water reaching inland to a distance of 1,000 steps or three gunshots. This would be about a half to one kilometre.
Floating debris was described also to have been caught in trees to the height of two halberds. The weapon famously associated with Medieval Swiss soldiers features an axe on the end of a long staff and measures about 2m.
Prof Anselmetti and colleagues have studied the collapsed sediments in the lake, and have simulated the likely tsunami waves they generated.
They can show how the waves move and how they can be reflected off opposite shores in a complex pattern.
"Those slopes that were triggered in 1601 - they will need quite a while to be 'recharged'," explained Prof Anselmetti.
"But there are other areas where we can see metres of soft sediment that are more or less ready to be released, so to speak. And we do geotechnical analysis in order to quantify just how much we need to shake these slopes in order to make them unstable.
"So, actually, we can predict to a certain degree what sort of earthquake is required to trigger which kinds of slopes, and because our numerical codes then allow us to calculate the resulting tsunami wave, we are somehow able to say which areas will be mostly affected."
The major qualification here is that the frequency of quakes of sufficient magnitude occurring in the region is small.
Those same collapsed sediments in Lake Lucerne can be used like ancient seismograms to time the reoccurrence of big tremors.
This reveals there were six destabilisation events in the past 15,000 years similar in size to 1601.
A recurrence interval of 1,000 to 2,000 years is too long for one generation of people to pay much attention to, concedes Prof Anselmetti. Nonetheless, the risks need to be quantified, he says.
"Somehow it is our duty as geologists to remind society that this is natural hazard - we may have tsunami waves, they occur rarely, but it is a natural hazard they should be aware of without making of course too much of a panic."
The summation of the research is being presented to the EGU meeting this week by the University of Bern's Michael Hilbe.
Jonathan.Amos-INTERNET@bbc.co.uk and follow me on Twitter: @BBCAmos