I don't have anything to add to the reams of text written about the ability scores, per se. Even if I somehow added more signal, there's already so much signal and noise that whatever I have to say probably won't make a difference. If you don't understand the implications of each ability score midway through your first campaign, RPGs might not be for you, anyway. Maybe it's already been done, but today I'm going to focus more on the interaction of ability scores -- between each other, and the game itself. There'll be some review as well, just to maintain perspective.
The obvious need for any class with melee in mind, as bonus damage alone makes a huge long-term difference. Question is, what's your Dexterity? Having high scores in both doesn't hurt, but high Strength often means you don't need Dex that much. If you're a warrior you can get that AC up with armor and other means. (In these days of high tech your DM had better be enforcing encumbrance rules. If you can read this blog, you have no excuse.) That said, if you plan to forgo Dex for armor you might want to beef up your Constitution because you won't be dodging many area-effect attacks.
The only other real drawback to high Strength is that its interdependence on Dex and Con puts severe pressure on the player to turn mental abilities into dump stats. It actually improved dramatically between AD&D2 and D&D3; the AD&D2 equivalent of a 3rd Ed. 12 Str was 17! I'm not a fan of stereotypes but it seems the only thing keeping every warrior from becoming a dick is the Paladin class. On a more tangible level, a low Wisdom score makes warriors particularly vulnerable to mind control, where all their destructive power gets turned on the rest of the party. Some players and DMs consider this "cheap" but I prefer it as a way of keeping character creation honest. If you put your three best scores in Str-Dex-Con, as a DM or a player I'm targeting your low Wisdom.
Dexterity's fundamental combat role (accuracy & dodging) hasn't changed since at least AD&D, which is why lighter classes like Wizard can never afford to overlook it -- it was one of two ability scores (the other being Con) that kept you alive. About the only thing keeping spellcasters from putting a high score in Dex (if they had one to spare) is that there was no statistical difference between a 7 and 14 until D&D3 came around. There are only so many 15+ scores unless you're blatantly cheating, and the second one always went to Con because Hit Die rules were downright brutal back then.
Dex has increased in importance over the years, largely thanks to efforts to make Rogues more relevant in combat. The modern skill system (first integrated into core rules in AD&D2) and feats like Weapon Finesse or Two-Weapon Fighting make it an attractive alternative to Strength for characters that aren't planning on brawling in armor, since its defensive benefits remain in place as well. However, the attack bonuses it provides don't mean much unless the attack has some pop. That limited its usefulness in earlier systems; in later ones, it's intended to work with ranged touch spells or sneak attacks. To make a difference in combat without a high Strength score, you need to have some other means of delivering damage.
To sum up, Dex doesn't interact with other abilities so much as your basic survival and offensive strategies. If you've got the Str to wear armor & dish out damage, and the Con to absorb punishment, Dex can be mediocre or even a dump stat. If you intend to use Dex as a means of enhancing your offensive and defensive abilities and you're NOT a D&D3-Pathfinder era rogue (about the only class tailor-made to convert Dex into offense), you'd better have a plan.
For such a fundamental and basic stat, Con is remarkably tricky. The reason is because it enhances survival and not much else -- it's not the most important stat for ANY class. D&D3-Pathfinder systems primarily use it as a capacity for raging, but you certainly didn't want to put your best stat in Con just for that when Strength was your go-to ability. Warriors typically don't skimp on Con, anyway.
Like Dex, your Con score should be determined by your overall survival strategy. I've made some PCs that effectively used whatever points that would've otherwise gone into Con to round out a "who needs hit points if the enemy dies first" strategy, but it's certainly risky. Sometimes this tactic gives the DM the impression the game is too easy without realizing I'm sweeping through fights because I can't risk staying in combat for too long. The DM adjusts, and my PC quickly goes from unstoppable to dead. Offense aside, if you've got decent AC and saving throws, you can probably ease up on the Con. (Note this excludes saves where hit points are a non-factor, such as vs. enchantment/charm spells.) If either your AC or your saving throws are weak, you should have a decent Con score. If your offense is bad, your AC is bad, your saves are bad and your Con is bad. . . uh, good luck.
Personally, my Con usually winds up being "mediocre" because I try to build characters around strategies solid enough to not need tons of hit points to survive (i.e. the high scores go to other abilities), yet not rely entirely on said strategy to survive because shit happens (so Con is never a dump stat either). When I didn't follow this advice and used Con as a dump stat, I went entire sessions without taking damage only to go from healthy to dead from a single (un)lucky blow. On the flip side, I recently tried a Fighter with a high Con but the guy was basically a wall with legs. It was effective but rather boring.
The importance of Intelligence depends largely on your DM's style. If the campaign is a glorified tabletop version of Diablo then Intelligence is only meaningful if you're a Wizard. That's not interesting to talk about, though, so from here on let's assume the DM expects players to respect the importance of skills and (in-game) social interaction.
The languages and skill points are important, and it's a prerequisite for a few nifty feats (see my earlier post on how I turned a low-Str, high-Int Fighter into a game-breaking monster using Pathfinder's Combat Maneuver rules). We're still looking at Intelligence in a bubble, though. The only class that has to make any sort of decision between Intelligence and another ability is the (D&D3-Pathfinder) Rogue. They get so many skill points that it's easy to focus on Dex (which is important for combat anyway), but what if you made a spy-like character? The Dex bonuses are in many of the wrong skills and you can compensate for the few agility-based skills you use with extra points.
As an aside, I had a couple house rules that made Intelligence more important. In particular, the number of seconds (IRL) you had to decide your action in combat was equal to your Int score. If you hesitated when declaring your action after that, you forfeited the round. This didn't really handicap anyone as even a warrior with 6 Int had six full seconds to start declaring the hack n' attack everyone saw coming anyway. If anything, it forced players of spellcasting classes to make their decisions faster, which was a good thing. If it had any sort of effect in line with its intent, it prevented players from trying to get clever with characters that clearly lacked the Intelligence to think their way through combat. You're a Barbarian, dude; stop fussing with alchemical items and start swinging!
Ah, Wisdom. Never underestimate the horrifying impact of a mind control spell. Players hate that but I don't accept the excuses because there's a very direct means of improving your resistance to those sorts of attacks. You don't get the superfreak athlete you were hoping for, but that's the difference between adventure and just an endless series of fights. Honestly, damage by weapon is only one way to die and in some campaigns not even the most common method.
Let's hammer this point home: If you're going to put a high score in Wisdom, odds are you're giving up a physical ability score because the non-Cleric classes that need high Wisdom are those with crappy resistance to mind-affecting attacks. However, the more you piss on mental stats to turn your warrior from a freak to a monster, the more devastating it'll be when you inevitably get your brain raped.
On that note, because classes like Paladin or Wizard are inherently good at resisting mind control, in these cases I often settle for a mediocre Wisdom in favor of Dex or Con. If this sounds formulaic, it's because adventuring is a hell of a form of natural selection.
Some DMs use Wisdom for ability checks to prevent players from making stupid decisions. If a PC is about to commit suicide by stupidity -- Wisdom check! Flipping that around, though, I will sometimes punish low Wisdom by demanding a check for decisions that make too much sense made by characters that are statistically foolish.
OK, so the conventional take on Charisma is that you can either be effective in combat or get laid. The Paladin seems to be an exception, but that's why as a DM I sometimes rule they're chaste. >:D
Pathfinder makes this ability more interesting because classes like Cleric and Bard also rely on Charisma, increasing its importance. I imagine most players see it as restrictive (especially Clerics) but I've tried throwing high Cha scores into Pathfinder characters and it works pretty well. Pathfinder aside, any post-AD&D2 Rogue can forgo Dex entirely in favor of an Int-Cha combination to make a social chameleon instead of a sneaker.
I like the Pathfinder changes, but like Intelligence, the importance of Charisma still really depends on how the campaign is run. Many groups understand Charisma well enough to designate one character the face of the party (the one character with a decent Charisma), but I think DMs are too forgiving here. One slick talker with a bunch of uncivilized brutes in tow shouldn't have unrestricted social privileges; when does that work in reality? I have to admit that most other classes won't make regular use of high Charisma, but if the PCs spend a lot of time around people and on infiltration missions the DM ought to inversely correlate the difficulty of the challenge with the party's minimum Charisma. If even the warrior's not a liability in social circles then the party should have the flexibility to pick and choose their battles. If even the median Charisma is below average then non-combat options should be very difficult to come by -- heck, the players probably prefer this anyway, given that's how they designed the characters. They just shouldn't have it both ways.
If there's a lesson here, it's that players will often use Charisma as a dump stat because both sides are stuck in a status quo where there's no incentive to have more than one high-Cha character because of how D&D is conventionally run. We gotta bust out of that pattern yo, because you don't show up to a high-class social gathering with a bunch of fly-ridden primates and get away with it just because you've got a nifty suit and a firm handshake.