Short tournaments Edit
It seems that the majority of smallish tournaments found in local casinos around the world these days are relatively short tournaments, intended to generate excitement and appeal more to the "gambler" than the skillful player. These tournaments are generally structured to run between 60 and 120 minutes in length; tournament levels run between 10 and 20 minutes, and the blinds increase rather steeply. Additionally, players are generally only given somewhere around 20 to 40 times the (initial) big blind in chips to start the tournament, and rebuys, if available, are few or small.
These tournaments run so fast and the blinds increase so rapidly in relation to your chipstack that winning players must play aggressively and generally much looser than they might typically wish. If you wait for a premium hand before entering a pot, you could bust out of the tournament before playing even a single hand.
For short tournaments, there are essentially no "early rounds": you start out in conditions you might consider the equivalent of "halfway through" of a more leisurely tournament. Thus, the general rule of early tournament play (patience and study) must be modified. The shorter your chipstack with respect to the blinds, the more you must make all-or-nothing decisions on hands. A general rule of thumb is that once your chipstack drops below 10 times the big blind, your most effective move will be to push all-in with any hand with a good chance to be best (often, the range of hands this includes is quite large: any ace, any suited king, any pair above fives, etc).
Long tournaments Edit
For long tournaments where you have a deep stack, tournaments where you could (if you desired to do so) fold hundreds of hands without playing before being blinded off, you have the freedom to be much more selective in your starting hand selection, and can play it more strategically and more like a cash game. Note that you should not play a tournament exactly like a cash game: there are important differences, not the last of which being the ever-increasing blinds that will "devalue" your remaining chips at a relentless pace.
There are two basic strategies for playing the early rounds of large tournaments (and, of course, many variations on these, as well as mixtures of the two):
Many players (amateur and professional alike) subscribe to an Aggressive tournament strategy and swear it is the best means to win a large or long tournament. The goal of an Aggressive strategy is to get involved in a few big pots early and win them, thus taking a commanding chip lead into the early rounds. This lets the player dominate the rest of the tournament. The idea is that by doing this, you can force all other players to "play catchup" to you the rest of the way.
The problem, of course, is that in order to do this, you must get involved in a number of big all-in pots early, and you will need luck to win them and get this massive chip lead. If you lose these pots, you are frequently either busted out of the tournament or severely crippled. Even if you only get all your money in with the best hand and even if in each pot you are a 2-to-1 favorite, if you have to win 3 of these pots to take the chip lead, you are a 2-to-1 dog to win all three.
Still, some players say this strategy is more effective and a better use of their time to enter a number of tournaments, and would rather have a 80% chance of busting out early (thereby giving them time to enter another tournament if they like!) if that gave them a 20% chance of becoming the huge chip leader. For tournaments where first prize is quite large (e.g. the WSOP), this strategy has some desireable elements: not only does becoming chip leader mean other players must "catch up" to you, it gives you a significantly better chance of winning the top prize, which could mean millions of dollars.
While this strategic approach is not called "aggressive", it is not intended to indicate a playing style which would be classified as "passive". When you play a hand, even in the Selective strategy, you must play it forcefully and aggressively. The difference is that your goal in the Selective strategy is to slowly build your chips throughout the tournament.
The benchmark chip measurement for Selective play is the "average chipstack", or the "median chipstack". Tournaments often make known one or both of these numbers during their course for you to use in determining how well you are doing. The basic goal of Selective strategy is to always remain above the average in chipstack.
Selective players will tend to play tighter and, more significantly, will tend to bet smaller amounts, working on winning small and medium pots and not getting involved in too many large pots: large pots always carry risk that another player could suck out (or have the best hand all the way), and large pots mean a large loss to your chipstack if you lose.
General Advice on Going All-In Edit
Going all-in during the early stages of a deep-stacked tournament is a very questionable play, because you'll fail to maximize value from your strong hands. Harrington on Hold'em volume 1 has a great discussion about why pushing QQ with 100+ BBL stacks is foolish.
Conservative play is good early in a tournament, because you can maximize your edge against your opponents once you have an idea how they play.
Some people object to the idea of staking all their chips on a single hand, even a nut hand such as AA preflop, early in a tournament. But in reality you'll need to double up several times to win a major tournament, and you'll never get a better opportunity than when you have the nuts. (In some variants, most notably Omaha, you might be even money heads-up with the current nuts, but you should never be much of an underdog.)