As I wrote before, I spend a lot of time lately to prepare my new Dungeons & Dragons 4th edition campaign. However I wouldn't go as far as saying that I am "planning" that campaign. The difference between the two is in how far I foresee what will happen. In my opinion, planning every detail of how the story will go is a bad idea, as it will only lead to clashes with the players once they inevitably stray from the planned path. It is far better to use an approach which is a lot less strict: Prepare the area, the NPCs (both allies and enemies), the history and interaction between those NPCs, and the locations. And then just let the story evolve by itself, depending on the decisions of the players.
That is especially important as most of the adventure I am planning is either outside or in a village, and I reduced the dungeon part of the adventure to not take up more than one third of it. For a dungeon preparing the next room and planning that the players will go there is practically the same thing. At best you can give the players the choice of whether to explore a room not on the "critical path" from entrance to final encounter. But in the wilderness or civilization players have a lot more choice. When presented with a new situation, how will the players react? If there is more than one simple and linear story thread, which thread will the players want to follow first?
Of course there is a danger in the other sort of extreme too: While a dungeon might be too linear, a completely open-ended story might give the players not enough direction to actually come to the end of it. And that is where the preparation comes in useful: If the DM knows how all of his NPCs are related to each of the story threads, every interaction the players have with the NPCs can be made to give some hints, even if that interaction comes from a player idea the DM didn't plan for.
In a pen & paper roleplaying game it is important that the players drive the story, and not the other way around. Nevertheless a DM can plan events that happen if the players *don't* do something. Thus the outcome of the story still depends on the players, but the adventure doesn't get stuck because the players for some reason didn't follow some particular story thread. The advantage is that such stories feel more alive than static dungeons, which appear frozen in time until the players open a door.
Thus my preparation consists of knowing what happened before the players even arrived, what each NPC knows and what he wants, what the players will find if they go to specific locations, and what would happen in the future if the players wouldn't intervene. I can't exactly plan that intervention by the players, but I must make sure that there are several ways in which they could get involved and can influence the story, so that the players never become passive spectators.
As I like writing (who would have guessed!) I do write down one possible sequence of events like a commercial adventure module. But that is all it is: One possible sequence, one that might happen if the players do what I would think is "obvious". Only that of course experience shows that what the DM thinks is "obvious" might be very much not so to the players, or that even if it is the players might decide to do something different. But with just a little preparation work on the background and motivation of my NPCs, it becomes easy to decide on the spot how they would react to some unforeseen idea of the players. And not knowing what exactly will happen until I actually play the adventure with the players ultimately is part of the fun.