For my Bleakstone  hex crawl, I decided I wanted a random encounter system that met the following criteria:
- Conforms roughly to the baseline 1/6 chance of an encounter/day in Basic D&D
- Can be weighted by terrain, region, PC activity, etc.
- Not entirely predictable by my players
I couldn’t find a system like that in any of my books, or online, so I built one.
My baseline was Labyrinth Lord’s system (which is the same as Moldvay Basic D&D, which LL clones): Roll d6 once a day, and there’s an encounter on a 1. That looked too simple and too predictable for my tastes, because it ignores terrain (etc.) and players can count on 0 or 1 encounter/day when traveling.
On the other end of the spectrum was AD&D 1e’s system, which involves up to 6 rolls/day at different times of day, is weighted by terrain type (forests are more dangerous than plains, etc.), and produces encounters on a 1 on a d10, d12, or d20 roll (depending on region type). That sounded a bit too fussy for me, and the time of day element has the potential to be predictable.
I like aspects of both of those systems, but I wanted to land somewhere between them.
The roll is d10+d10, and a modifier of +1 to +5 can be applied depending on terrain type, region, or other factors (wagon trains, 30 mercenaries in tow, etc.). d10+d10 gives a pyramid-shaped distribution, and out of the various rolls I tried it was the easiest to map out.
Below are three tables showing how this works in practice: d10+d10 at the top, d10+d10+3 in the middle, and d10+d10+5 at the bottom. (You could go higher than +5, but +5 gives a 50/50 chance of 1+ encounters, and that’s where I wanted to stop.)
The middle — a forest, maybe — corresponds roughly to a 2/6 chance of having an encounter, but there’s a 10% chance of two encounters. And the bottom is the most dangerous region, the Haunted Moor of Doom, wherein there’s a 3/6 chance of having at 1+ encounters and a possibility (6% chance) of having as many as three.
I like that this system has a small footprint (in my rules document, it’s just one table, the first one; the other two are for illustrative purposes), scales easily based on whatever factors I want it to, isn’t immediately predictable but also isn’t wildly unpredictable (players should be able to catch the increased frequency of encounters in some areas, but not know for sure what the cap is), and aggregates what could be multiple rolls into a single roll.
Time of day
If I need to randomize when an encounter will occur, that’s as simple as rolling d6 for each encounter: 1 Morning, 2 Noon, 3 Evening, 4 Night, 5 Midnight, 6 Pre-Dawn. (Those are the same times AD&D 1e uses.)
If you want to see the math behind this system, check out AnyDice , my favorite online probability calculator. Here are the results for d10+d10, +3, and +5 . My thanks to our own Matt Neagley  — who’s a probability wizard — for helping me crunch and verify the numbers, presenting an alternate system, and talking me through a lot of the math.
If this system is useful to you, or looks horribly flawed in some way, sound off in the comments!