I mix and match my own stuff and pre-published; and also get adventures from a third source - stuff my friends have written.
I'll usually have some sort of plotline at least vaguely storyboarded out in terms of what adventure follows what, and if I find I can shoehorn a canned module in there that's great; as it's one less I have to dream up. But sometimes I just have a module I want to run, so I drop it in anyway whether it fits the story or not (there have to be *some* red herrings, don't there?).
And sometimes I'll just take a canned module or series and butcher it into unrecognizability, then run it. A recent example:
I wanted to run the Slavers series A1-A4. But just before I started, another DM (with whom I share players) decided to run A1 by itself, so I had to improvise. I dreamed up a two-adventure substitute for A1, then ran A2 pretty much stock, ran A3 but nowhere near the way the writers intended, and bailed on A4 as I don't want Suderham destroyed quite yet (I have plans for it later). They then went back to Suderham and got captured which let me run a homebrew variant on A4 anyway.
A long series I hope I'm just starting - in other words, if the party don't bail on it like they did the last series I tried them on - will go:
1. canned (0e)
2. canned (4e, very modified and with lots of homebrew elements around it)
3. homebrew (new)
4. canned (1e)
5. homebrew (barely touched in a previous campaign thus I can recycle it)
6. canned (1e)
Another series I'm just finishing went:
1. homebrew (new, inspired by a song)
2. homebrew (new, but partly inspired by an old adventure a friend wrote for his game)
3. homebrew (new)
4. canned (1e, and not planned at series start; it just happens to fit so well!)
Lan-"I love it when a plan that isn't a plan comes together"-efan