The simple answer is that it is transparent and easy to understand. Every one can interpret their chances if they need to roll 42 or less on a D100. It's simple and elegant.
Then there is also my gaming history. As I explained in an earlier post my first game that I played a lot was RuneQuest II. D100 became the golden standard for rpg mechanics for me.
I know that there are more interesting dice mechanics in later rpgs. I really like both Greg Gorden's in Earthdawn and Jay Little's in Warhammer Fantasy Roleplay 3rd edition. But the whole idea with my game is to be old school, so we will use a straight D100 as the main resolution mechanic in the game, and if I would construct a more contemporary game I would probably use cards as the resolution mechanic as cards are highly customizable (but that's another blog).
There is an interesting discussion on reddit under the heading What's the problem with percentile systems. The discussion, I think, misses the distinction between using the 1-100 scale for skills and the use of a D100 for skill resolution. They can of course be used together and you can use a D100 for rolls without using the 1-100 scale for skills (maybe not the other way around, though). I understand that people have problems with a 1-100 scale when you need to add and subtract values or when you use a system like RoleMaster where you add (or subtracts) D100 rolls from each other. The math isn't complicated but it slows down the game. I think a 1-20 skill range combined with a D100 resolution system is the way to go (as we will see).
It would be possible to build a system that only uses D100 for all game mechanics, but much of the old school fun was to sift through your dice bag to find the D8 or that special shiny D4. Weapon damages and randomized spell effects could all be accomplished with all those other dice D4, D6, D8, D10, D12 and D20. A bit of variation from the main D100 mechanic.