No one really knows. The only time it has been managed was once and that relied on having a tracking device on the ICBM, on it's source, path and target being know before time and on it following the simplest possible trajectory. And that's assuming the Koreans only launch 1, presumably if this thing works in June, they will make more than 1...

It's surprising how many systems there are (Russian, French, British, US, Chinese, Indian etc) and how rarely anyone dares to actually do a public test...

Single ICBMs can absolutely be intercepted, but the easiest they are to intercept during the launch phase where they are still accelerating, aka close to the launch site. An ICBM in its final phase of flight that's falling towards some random city... way harder. But yes, the nuclear threat relies on there being multiple ICBMs, with decoy warheads that are also launched so that you never know which one is the real bomb.

Definitely can't be intercepted if you completely go around the interceptors—launching north, going around the world once, and hitting North America from the south. That's the fractional-orbit [0] concept the parent comment alludes to.

If North Korea were to attack the US in the most cost-effective way – crossing the Pacific east – there's GMD interceptors in Alaska that could try to hit them in space, several thousand km's above the ocean. It's unclear how well those work. The *public* version of the story is that they have about a 60% success rate, against simple targets without countermeasures; and that they intend to fire four interceptors against each incoming warhead to boost this probability [1].

[0] https://en.wikipedia.org/wiki/Fractional_Orbital_Bombardment...

[1] https://www.economist.com/united-states/2018/01/11/there-is-...

I assume shooting it down would potentially scatter debris over Japan.