In order to understand how the game has changed from its original concept, one has to research the rules of later systems and the changes made. It sometimes reveals interesting things about the original version that may or may not be obvious when seen only through the rules as they are initially interpreted without a comparison. I'll use the 2nd Edition rules (originally printed in 1989) to show how the rules evolved from confusion towards clarity.
I expected to find the surprise rules in the combat chapter of the Player's Handbook, but instead found them in Chapter 11: Encounters, two chapters AFTER the combat rules. There are also a few paragraphs and a chart in the Dungeon Master's Guide. The important thing to remember through all this is that the round structure changed from 1st to 2nd Edition. In AD&D 1E, the one minute round was broken down into 10 segments of 6 seconds each. Casting times, speed factors, and segment movement were all tied to this structure in some way. In AD&D 2E, there is no breakdown of the round at all. A melee round is still 1 minute long, but it is now the smallest unit of time in the game. The next important thing to know is that the die used for surprise changed from a d6 (which were easier to locate for AD&D 1E players) to a d10 (which, by the time 2E came along, polyhedral dice could be found virtually anywhere). I have learned that Gygax also used a d10 for Initiative in his own campaigns, but that has no bearing on this discussion (it's simply interesting to note that the game was evolving even after the AD&D rules were developed).
OK, so the rules begin on page 111 in my copy of the PHB. Not all printings have the same page structure so it may be different in your book. Mine is a 1st printing, so if your page structure differs, or the rules were amended at some point with corrections, then you can disregard any statements that may differ from your rulebook.
The very first line seems to indicate exactly what I've been stressing in my other surprise articles: surprise is usually unilateral. In this version, the chance is 3 in 10 (or 30%) which is also very close to the 33-1/3% of rolling a 2 in 6 (probably intentionally done). So far the rules for both editions support each other. It then gives examples as to how surprise can or cannot be avoided. Again, it stresses that it's the DM's call as to whether surprise conditions exist, and it depends entirely on the situation. Curiously, it says that the DM can have the group check or require individual checks. If no one is paying particular attention to something, saying they are alert for danger, or are otherwise distracted, then they can roll as a group. But say that 2 characters are standing watch at night while the others are asleep, then only those 2 watchmen roll for surprise, and in this case it would be each individual (thus the more watchmen that are alert, the less chance that the entire party is taken by surprise).
Now the surprise roll can be modified by Dexterity (reaction adjustment), race (elves and halflings), class (rangers and thieves), cleverness, and situation. The modifiers are given in the DMG, p. 102 (mine is a 10th printing to replace a badly damage 1st printing). Here it explains that surprise may be possible for BOTH sides. Light, excessive noise, and other types of prior warning can negate the need for a surprise roll, and surprise is not usually possible when no form of concealment is present (such as two ships at sea), but darkness, storms, fog, and the like can act as concealment.
It is important to note that the modifiers listed in the DMG apply to the DIE ROLL, not the chance to surprise. Thus, if a monster is silent (-2) and invisible (-2), the party's die roll is reduced by 4. If they rolled a 6 on a d10 they would be surprised since 6 - 4 = 2 (normal surprise chance is 3 in 10, remember?). If the party comes near a patrol of 12 troglodytes (+1 per 10 in group) who are camouflaged (-3) in a dimly lit tunnel (-1), the DM is within reason to give the party a bonus for the distinctive stench of the trogs (+2) and an additional +2 because a small band of fleeing, wounded orcs encountered earlier warned the party of the presence of trogs ahead. Yes, this can become very convoluted! However, the 3 in 10 chance remains the same regardless, and the die roll is modified by (+1 - 3 - 1 + 2 + 2 =) +1. So long as the party rolls a 3 or above they will not be surprised. Now, if the same party had never encountered the wounded orcs, and the trogs had not previously battled the orcs and released their stench, then they would have a modifier of -3 to their surprise roll, thus doubling the chance of being surprised! [NOTE: I'm using stats for monsters from the Monstrous Compendium binder pages printed in 1989. Those using the 2E Monstrous Manual might see that the chameleon power of trogs actually grants a stated -4 to their opponents' surprise rolls.]
The decision to roll group vs. individual surprise is the DM's call. I would just adjust the group roll for each individual's modifiers. Say that the group surprise roll was a 4 with no other modifiers to the roll. A character with a 5 Dexterity (-1 Reaction Adj.) would be surprised, while a character with an average Dexterity or more would not be surprised. Elves, when alone or well in advance of a party and not in metal armor, impose a penalty of -4 to their opponent's surprise rolls (or -2 if they have to open a door or screen to attack). This is different than 1E where elves surprise 66-2/3% (4 in 6) or standard chances (33-1/3% or 2 in 6) when opening a door. Granted, 70% and 66.6666% is close, but 50% chance is much more than 33.3333%. Elves also tend to have better Dex scores, so they would be less likely to be surprised by others. Halflings have the same modifiers as Elves. Half-elves do not share this ability in either edition of the game. Rangers in 2E no longer have surprise modifiers built into the class. However, they do gain access to Hide in Shadows and Move Silently (as do Thieves), so they benefit from the modifiers to the roll given in the DMG for being silent (-2) or invisible (-2). However, unlike elves and halflings, they cannot do both at the same time. A halfling thief probably DOESN'T gain the benefit of both his racial sneakiness AND his Move Silently check as these are both similarly explained in the rules. Therefore a halfling thief CANNOT surprise opponents 9 in 10, but CAN surprise 7 in 10! A human thief using his Move Silently successfully can only surprise 5 in 10 by comparison (which is equal to the 3 in 6 chance for a 1E human thief). And, of course, surprise is a requirement for a backstab attack to get the multiplier to damage and bonus to hit.
The other modifiers listed in the DMG seem to be more situational and deal with party size, lighting conditions, morale, concealing weather conditions, and anticipation of attack. These are all common sense rules. However, many creatures in the 1E AD&D Monster Manual had increased chances for surprise. Let's see about 2E monsters. Troglodytes are camouflage attackers, so one would think a modifier would be nailed down on these creatures, but the Monstrous Compendium only states that they have camouflage ability, and in the DMG camouflage is given a range of -1 to -3 (which oddly enough can equal or EXCEED being invisible, which is only -2). Bugbears, another traditionally stealthy monster, are specifically listed as surprising with a -3 penalty to their opponent's surprise roll (which is better than simply silent movement at -2). This is an improvement over 1E AD&D. Finally, the accursed Drow are STILL listed as surprising 75% of the time (can't be rolled on a d10 guys) and also move silently (which is apparently already factored into their elven cloak and boots). It states that they are only themselves surprised 1 in 10 (to which standard modifiers are likely to be added). Unfortunately, instead of converting from 1E AD&D percentages to d10 increments, the authors kept the weird, non-standard surprise chance of the dark elves. But this doesn't matter much (as will be seen) and each +/-1 of modifiers adjust the chance by 10% so it's just a math exercise.
There is a distinction drawn between surprise and an ambush in the last paragraph of PHB, p. 111. Here it states that an ambush is prepared by one group to make an unexpected attack on another group but ONLY if the DM decides that the other group CANNOT detect the ambush. A properly set ambush allows the use of spells and attacks before the other side reacts. If the ambush succeeds, the ambushing group gets its initial attack AND the other group must roll for surprise as well in the following round, so an ambushing group may gain TWO rounds of attacks before the other group can reply. This sounds suspiciously like SURPRISE and COMPLETE SURPRISE from Original D&D and AD&D 1E to me...
Now, what are the effects of surprise in 2E? We all know the confusing rules in 1E may result in one or more segments of surprise actions for the surprising group. However, there are no segments in 2E. Instead, all surprised opponents suffer the same penalty -- one round of attacks by melee, missile, or magical items only (no spells can be cast during a surprise round!). This means that all the fiddly surprise rules are basically out the window in 2E! You don't have to worry about movement rates in segments, casting times for spells, or anything else. If you can do it in a round, then you can do it in a surprise round, with the only exception being that spells cannot be cast. Although one would see this as being very unfair for magic-users, they can still use magic items (if available) and prevents instant-cast magic missiles and charm person spells from overtaking the party. It also allows fighters their multiple attacks with melee weapons, and any class to make use of standard rates of fire for missile weapons. Surprise also means that the surprised characters lose their Dex bonus to AC (and the term "flat-footed" here would be used later in 3rd edition for any character before they act in the 1st round of a combat). Mutual surprise conditions simply mean that one round is wasted for both sides -- since there isn't a segment comparison, all those rules from 1E are essentially unnecessary.
So 2E surprise tried to become normalized and streamlined for all monsters and characters (except, of course, those drow troublemakers). The 3 in 10 chance equates to the 1E chance for normal surprise, and complete surprise in Original D&D (2 segments) is now reserved only for ambush situations (which are DM arbitrated and not left to random roll). Removing spells from the surprise equation and making surprise last a full round eliminates all the headaches of segment calculations. Of course, there is still complexity involved in the amount of modifiers that have to be remembered for each circumstance, but as long as the DM and players have a grasp on their characters' abilities it shouldn't be too hard to figure out. It's also interesting to note the slow transition of the surprise roll into the Perception check of 3rd and later editions.
