I marked off 0-1 hours but that's a bit misleading. I typically do an obscene amount of work before a campaign starts following in line with my own "Prep to Wing It" approach towards the game. Stat up major NPC's, get statblocks for throw-away fights (Generic_Assassin_01, Bunch_O_Orcs, Guard_53, etc), and sketch out the timelines for all the major current events. The list goes on.
*Most* sessions will be little more than refreshing notes, looking over all the threads the PC's are pulling (or are pulling them) at the moment, updating timelines and Who's Who When and Where lists. Making sure everything in my head is synched up with everything on paper. Nearly all encounters are triggered by things the group does. Prep time only goes above an hour when I want to put that much time in or when I know I'll need to (PC's planning on raiding a house for documents requires a floorplan, guard rotations, a few quick statblocks, checking to see if anybody important is there at the time, potential traps or surprises, and laying out any additional plot hooks found while inside, etc etc).
I imagine the fact that I only make battle maps for very important or difficult to explain/visualize fights cuts down on that considerably. Random fights were generally sketched on paper or a dry erase board to keep distancing, environment, and visuals - the fights then took place organically, rather than statically at X intersection or Y clearing. I grew up on AD&D without battlemaps and have never had a problem visualizing, nor my players, even with 3.x. I tried it out for a few months right when 3.x came out but I found that it unconsciously led me to guide towards encounters I prepped much more than I was comfortable with, as well as muchly increasing prep time, and when the players started asking why the game felt tighter than it usually did that's when I stopped using them again.