Do you mean the actual battle calc? What annoys me is when I smash through dozens of units over and over again (In any game like this), and then one of those dozens gets a lucky hit in for massive damage. Had it happen in GalCiv2 where a standard stack I had killed dozens of times attacked, the enemy ships consistently dealing 0 damage, and then one of them randomly does 50 damage, one-shotting one of my ships.
Yep, that's a inherent problem in the 1-roll, winner takes all type of combats. While overall the odds are the same, the variance is much higher in that kind of system. A better system is a multi-roll simulation where it takes multiple battle rounds to resolve a battle and one side comes away damanged.
Most games, including GalCiv2 and Civ4, have gone the multi-roll route actually. How far they've gone I don't know, but in both you rarely escape a battle as the victor without some damage or potential for damage. Still, as long as dice are being rolled, you can get oddities in the distribution. Having played enough poker, god knows I've seen quite a few 1000-1 things happen (well, 1/23*1/45).
With the massive armies Brad has mentioned I think the variance problem may be reduced somewhat. If you've got 1000 vs 1000 infantry being modelled and you're really rolling 1000 times, plus doing it over a few rounds of battle, then the odds of a freak event happening go way down. Civ4, GalCiv, all of those tend to have very few units at any given time, so even if you do multiple battle rounds you're talking maybe a dozen rolls, not thousands.