Let me suggest an analogy.
Suppose you take a mole, $6 \times 10^{23}$, of coins, all head up, and you start flipping them. Any coin that comes down tails is discarded. After the first flip we'll lose (about) 50% of the coins. after the second flip only 25% will be left, after the third flip only 12.5% and so on. So we'll start losing coins very rapidly, but because we had so many coins there's a fair chance at least one coin will still be there after 79 flips (because $2^{79} = 6 \times 10^{23}$).
Was the coin that came up heads 79 times in a row any different to the other coins? No, it was just pure chance. Could we have predicted which coin was going to survive 79 flips? No, because all the coins are identical.
This is what is going on with your radioactive decay. We start with a mole of atoms, all identical, and every atom has a probability to decay within one half life. Half the atoms will decay in one half life and half will be left. two half lives later 25% will be left and so on, just like the coins.
And just like the coins the atoms that manage to survive many half lives are no different to all the other atoms. It's just pure chance that some survive a long time and some don't.
Response to comment:
Avik makes the good point that I've assumed all the coin flips are random. In principle they aren't, since if we knew exactly how the force was applied to the coin we could calculate it's trajectory and predict which side it would land on. This may be hard to do in practice, but in principle it's possible.
In the case of radioactive decay the nucleus has a lower energy state available (the decayed state) but there is a potential barrier that it has to climb over to decay. The nucleus doesn't have enough energy to climb the barrier, but it can get through the barrier by quantum tunnelling. Unlike the coins, quantum tunnelling is (as far as we know) a truly random process. That means no matter how carefully we measure the state of the nucleus we will never be able to predict whether it will tunnel through the barrier or not in some pre-determined time.
I should also mention, because it's all part of the same physics, that the uncertainty principle means we can never measure the state of a nucleus precisely anyway.