The goal is to require some form of trade-off with every decision. One example of this is choosing between a smaller but safer bonus, and a riskier but much more powerful one.
This generally works best if the safer bonus is safe simply because it pays off in the short term while the other option pays off later and is riskier because other factors can come into play.
If the choice is basically just between "25 percent chance something really good happens" and "75 percent chance something okay happens" you're not really making a strategic choice as much as you're gambling.
While this can be made to work, it's tough to make an interesting decision out of swinging a large weapon that only hits 50 percent of the time but does 50 damage or using one that hits 90 percent of the time and does 25 damage.
More often than not players will go with the safer option (especially if they try using the big one and it misses three times in a row -- good luck getting them to try using it again!) This sort of mechanic is fairly rare in modern games, but I still see it pop up in a few Japanese RPGs.
You can find good examples of short term safety versus longterm risk working really well in pretty much any sports team management sim where there is a longterm player development aspect or the risk of injury/underperformance.
I've been playing a lot of the text-based sim Out of the Park Baseball for the last six months or so, and there have been some agonizing decisions -- should I trade this guy in his prime for a collection of younger, less-developed and much riskier prospects? Do I send off a package of two risky prospects in exchange for one safer one? The inability to predict the future coupled with an understanding of the game's mechanics, the likelihood of certain types of prospects panning out, the players' injury history, the value of certain types of positions, etc. all combine to provide a set of very interesting and very difficult decisions.
This sort of trade-off can be applied in nearly any game. Should I declare war on my neighbor, hoping that the resources I expend to capture their lands add up to less than what I stand to gain? Is it worth the risk to attack that optional boss and try to open up a new area to explore? Players need to be equipped with some measure of information for this to be possible, but don't underestimate the positive impact imperfect knowledge can provide.
I know some people disagree with me on this topic and claim that a game needs to be "more than just numbers." You won't find me arguing against the value flavor and "feel" provide. There are an innumerable number of valid design approaches that can result in a game enjoyed by a large audience. The point is simply to recognize that when it comes to game mechanics there's basically a scale that has strategy at one end and flavor at the other.
One great example of a game which definitely doesn't go out of its way to share everything with the player is Dwarf Fortress, which is all about exploring the game space and laughing as all sorts of crazy things happen. While most players don't like there being a risk of drinking a potion and the game permanently ending right then and there, there are absolutely some who do and we developers definitely shouldn't write them off. The roguelike genre in general is characterized by this possibility of a single mistake derailing everything.
Every development team has to decide for itself where on the spectrum of strategy-vs-flavor it wants to land. What is the goal of the game? Who is our target audience? What do we want players to feel when certain events happen? No gamer is the same and no game should be either.
A point I often make when discussing game design is that the only manner in which a game really matters is inside the player's head. You could have the coolest, most complex system modeling some really interesting phenomenon... and it's completely irrelevant unless the player knows what's going on and how to have fun with the situation. When someone understands the mechanics and the implications of their decisions and is able to translate that into a completely unique experience -- that's when a game really succeeds.