For an asteroid ten kilometers or larger, it’s effectively zero. The automated searches for asteroids by Pan STARRS etc have been very effective. They have found all the asteroids that do regular flybys of Earth of ten kilometers or larger. The largest known potentially hazardous asteroid is Toutatis at diameter of 5.4 km but it won’t hit us in the next century. There is nothing larger than 10 km that can hit us at all at present.

Astronomers can use those observations to work out their orbits for centuries into the future. None of the known asteroids will hit Earth in this century. You can check this by going to the Current Impact Risks Sentry Table which shows the combined results of astronomers in dozens of countries world wide. If the top entry is white, blue or green there is no significant chance at all of an impact. The blue ones are of diameter 50 meters or smaller. They could cause local damage - that’s like Chelyabinsk, but no global damage.

A few of the flybys we get are by comets, about 1 in 146. So there’s a tiny chance of a comet impact by a 10 km comet in the next century. But the chance of an asteroid impact is 1 in a million per century. So for a comet that’s 1 in 146 million of an impact in the next century. That’s so tiny we can forget it.

Sometimes Earth may get more comets than usual and then the comet impact risk may be significant. But at present it is tiny.

We get asteroids about 1.5 km in diameter or larger every million years. So one chance in 10,000 per century.

Meteor Crater in Arizona was caused by a 60 m iron meteorite. We get impacts that large perhaps once every 1000 years so 1 in 10 for the next century. But most will fall in the sea or in a desert. It’s not large enough to cause a tsunami.

Even a large asteroid of 1 km in diameter if it lands in the sea may not cause a tsunami. The reason is that the sea bounces back from the impact - unlike an earthquake induced tsunami where the entire sea bed below the origin of the tsunami is uplifted. So the waves from the impact will probably rapidly diminish in height and it won’t have effects except locally. It’s large enough to have some global effects such as some global cooling, like a large volcanic eruption.

However we have now found more than 90% of the 1 km and larger asteroids and expect to reach 99% by the mid to late 2020s. So the chance of one of those this century is already reduced to perhaps 1 in 100,000 and if we find 99% and still none are headed our way, then its a 1 in a million chance.

We could find out for sure by putting some more money into asteroid detection. The B612 foundation used to recommend Sentinel, a space telescope to orbit just outside the orbit of Venus, at a cost of $500 million to look outward with infrared. They now propose instead a fleet of eight cubesats using synthetic tracking technology. This would cost only $50 million and find most of even very small asteroids that do regular flybys within 6.5 years.

. The idea is explained in techy detail in this paper. Finding Very Small Near-Earth Asteroids using Synthetic Tracking. For an easier to read summary of it, see “Synthetic Tracking” Set to Revolutionise Near-Earth Asteroid Discovery

The idea is that instead of doing a 30 second exposure, you do many shorter 2 second exposures. With conventional CCD's that adds to the read noise so you get more errors but there are new CCD's developed for medical imaging that permit fast accurate reading, called Scientific CMOS detectors. The Andor Zyla is an example here.

Andor Zyla 5.5 | sCMOS Camera medical imaging camera capable of fast read out with low read error

You can then use this to simulate tracking the asteroid with the camera, which makes the asteroid far brighter in the images.

This image shows a the result of stacking many photographs of asteroid 2009BL with camera set to follow the stars on the left - notice how the asteroid is shown as a streak, and rather faint. On the right, the same photos are stacked to follow the asteroid which then shows as a much brighter spot, and the stars are streaked and fainter.

Image from: DETECTION OF A FAINT FAST-MOVING NEAR-EARTH ASTEROID USING THE SYNTHETIC TRACKING TECHNIQUE

When the asteroid is small and traveling faster across the field of view, the trail can be so faint it can’t be distinguished from background noise when the camera follows the stars. If you know its velocity you can make it much brighter by following the asteroid. But what can you do if you haven’t detected it yet and don’t know which way it is moving? The idea of synthetic tracking is that you take lots of short exposure photos and just try stacking them in many different ways until you find the right velocity and an asteroid pops into vie win the photo. This is time consuming but modern graphics cards permit fast parallel processing which makes synthetic tracking feasible.

This approach can make it easier to spot fainter asteroids. It might mean for instance that you can spot an asteroid ten times further away than before. That means a thousand times the volume of space covered. So this technique can lead to a huge increase in the detection of asteroids.

We could retire most of the asteroid impact risk for $50 million with this new technique

The researchers found that fewer than eight cubesats, fitted with 15 centimeter synthetic tracking telescopes could find more than 70% of NEO's larger than 45 meters in diameter in less than six years (these are the asteroids that are most hazardous for us). The total cost would be $50 million. With larger 30 cm telescopes then eight satellites could find 95% of the NEOs larger than 45 meters in diameter in the same time period of less than six years. For details see their 2016 Annual Progress Report.

If I was a president of the US (not that I’d want to be or be capable of being :) ), my first executive order would probably be to set aside $50 million to build and launch that fleet of eight cubesats. Surely Congress would pass that law if it was well argued for. Just about any developed country could find that amount from small change in their defense budget. They could unilaterally retire pretty much the entire threat from asteroid impacts.

Many individuals who win the lottery could also find such an amount, as well as individuals who are multimillionaires or billionaires through inheritance or from their own work.

If we knew the orbits of these asteroids decades before an impact, they would be easy to deflect. It only takes the tiniest of nudges to deflect an asteroid if it impacts on Earth several decades into the future while it takes a big push to deflect it given only a few months of warning.

Some people have been killed by meteorites. But it is very very rare. Including two reindeer herders in Siberia killed by the Tunguska impact. See my answer to What are the chances of an asteroid or comet hitting the U.S. in the next 50 years?

$50 million spread over the US population, say, is a one off cost of 15 cents per person. In the UK (where I live) we could do it for a one off cost of £1.60 per person - that's to find them for the entire world. As an example, the UK recently voted to renew Trident at a cost variously estimated at £40 billion to £205 billion ($50 billion to $256 billion). The cost of finding 70% of Near Earth Asteroids down to 45 meters within six years is a tiny 0.1% of that. We can certainly afford to do that.

See also my Giant Asteroid Headed Your Way? - How We Can Detect And Deflect Them