Resources are used both for production and research, so one of the most important things in the game is to keep getting them as fast as possible. As most of the maps have 7-8 minerals, it isn't enough to put just one worker (SCV, drone, probe) on it because you'll very soon begin to lack resources. On the other hand, if you build up twice as many workers as you have minerals and put them to work, then you'll waste far too much resources for their production. So, the best way to gather minerals would be to put 3 workers on 2 minerals. This means that you must have about 10-14 workers for gathering minerals at all times. The worst way for gathering minerals I heard so far is to never stop building workers. That is, one guy said to me that he always lack resources even though he is constantly cranking workers. That was his mistake - not that he only needs minerals for the workers, but for the control also. For gathering gas, you should put 3 or 4 workers on it, depending on the distance of the geyser. Lately I figured that number of workers you need does not need to be calculated by number of minerals you have but, in every game you should simply have 12 workers on minerals.
As you are now aware of the optimal number of workers you need at your base, the second question is how and when to get to that number. Most of the top players decide to constantly crank workers and when the sufficient resources come, then develop.
The best way to get to optimal number of SCVS for a Terran player is to build supply at 8 control. Following is barracks at 11 control and another when you get enough SCVS.
With Zerg build overlord at 8 control, do the extractor trick (build extractor, then a drone and then cancel the extractor) and keep pumping drones until you get either 150 or 300 minerals.
When playing with Protoss, usually build pylon at 8 control and warp gateway at 11.
This may seem like a slow way for each race to get attacking units, but when you start cranking masses of them, you will be unstoppable.
If you are playing with the Terran, you must be fast in scouting (especially in Brood War). I recommend using two of your SCVS in the beginning (usually the ones that build the barracks and the first supply) to find out where your enemy is and also to check out his initial plans (if any). As you get academy, build comsat as soon as possible, and then hot key it (my favorite is CTRL+0) to have it accessible at all times. If it's energy reaches 150 use it to scan out the map and/or your enemy's bases. When you get starport, research cloaking for wraiths. With it, they are a powerful scouting tool (both fast and with cloaking ability).
Many people use just their overlord(s) on the beginning to scout out the map. This can sometimes be tricky, however, your overlord can get killed, then you can get in deep trouble by not being able to produce any kind of units. This mainly happens when playing a Terran player, because both Zerg and Protoss need time to get anti air units. So, use one or two of your drones (after you have two overlords) to find out where your opponent is. My recommendation is to research burrowing in your 2nd hatchery when you get time (and resources) and bury one of your zerglings in every expansion place and in front of your opponent's base. As soon as you get lair and queens nest, build a few queens an throw parasites on your enemy's workers and some stronger units (i also like throwing them on the creatures). When the game gets under way, constantly use your scourges to patrol the skies.
As for the Protoss, I recommend using one or (if it is a bigger map) two of your probes to scout out the map (usually the one that warp in the gateway and the second pylon). Later, when you get observers, your trouble with scouting is over, but in some rare cases, observers won't do. Therefore I recommend making hallucinations of some of the creatures and with them completely scan out any of your enemy's bases without fear of being killed. You can also use mind control for this, but it's much better to leave it for your opponent's units.
Good unit control is only available with units that have ranged attacks. Let's set an example two groups of 4 hydralisks equally upgraded. You probably think that the luckier player will win this one. But, with a tiny bit of micromanagment you can win this without any, or with very little losses. Say, the one player just sends in his hydralisks in without any more control, you just need to shoot his hydralisks one by one, and you'll win. If your opponent do the same, you just retreat the unit that is being shot and his hydralisks will start chasing it while your other 3 hit some extra shoots, also you'll win. If your opponent starts retreating units like I just said, you must quickly order yours to attack a different target and when the running hydralisk returns, shot to kill him (this can be very tricky if the network is set to higher latency). Then, if you think that your opponent will do the same, select 3 of your hydralisks and attack his units one by one while the 4th hydralisk is set to attacking a different unit - this will make your opponent think that you just send your hydralisks in without any more control and he will most probably shoot your units one by one, and you will select that hydralisk and circle around the other 3. If you do this properly, you should kill all of his units without any losses.
By my opinion, macromanagment is like looking at the whole picture. Meaning that it is a ability to do multiple things at the same time. Many people, when attacking, completely divert their attention to it, completely forgetting to unit production, research and expansion. If the attack shows unsuccessful, they'll have allot of money but no/very little units. Before attacking, I usually use time to build up an expansion and fortify it a little, until my troops reach their destination. As an attack can take a longer time, I usually queue up 5 units to build up, or, if I'm Zerg, use time to upgrade as much as I can, while building some creep colonies for defense. You must note that not all of your attacks will be successful, as a matter a fact, most of your attacks will be futile, so you must plan what to do after it. This was just an example. Consider that if you're attacking and/or attacked that you must research and produce more units at the same time. If you master macromanagment you'll become a much better player.
Opposite of the macromanagment is the micromanagment. This is, also by my opinion, a skill that you learn with time and practice that allows you to get maximum of every action you do and/or every unit you use. Just before, I mentioned unit control as one part of Micromanagment and also it's the one that is used most frequently. A lot of micromanagment is used when placing spells, especially psi-storm and irradiate. If you put an irradiate on an unit you will surely kill it and damage a few other that were nearby, but with a little bit of micromanagment, placing two irradiates on different units you can kill a great deal of them in the same time. The same thing is with psi-storm, just a little easier. If you see a army advancing to your position, by placing a psi-storm on the fist unit you will either kill them all or damage them severely. When using every unit, be sure to get maximum of it, like when scouting you can use your worker to prevent your enemy's workers to build structures or kill them while building up (if Terran) or even you can try constructing a refinery on their geyser as they wont be able to kill it with workers. Also, when placing burrowed zerglings or spider mines on expansion places you can put them on the place where a base should be built, forcing your opponent to bring in a detector unit. Loosing a vulture without placing all three spider mines, battlecruisers without using yamato, corsair without using a disruption web is considered very stupid as you do spend more resources to get that abilities. There are so much examples for the micromanagment which you will learn eventually. Micromanagment rarely takes little time so be sure to mix up a good combination of micromanagment and macromanagment in every game because that is the only way to win against better players.
Probably the first thing that will come to your mind when you begin to play starcraft online is what race to choose. Of course, there is no perfect race in starcraft as it is one of the most balanced games made to date.
Good thing about Terran is their powerful defense all trough the game. Therefore a common Terran mistake is Turtling - closing up inside one base until the minerals are depleted and then seeking for expansion. A good player can easily overpower this by taking 3-4 expansion and developing mass of advanced units. Another Terran mistake is massing marines while not developing much. Simple counter for this are lurkers or dark templars.
Zerg players, on the other hand, tend to make rush as soon as they can. Most of the time, the simple 6 zergling rush will kill a uncarry player, but that's just the thing that spoils the game. Other common mistake is making tons of sunken colonies and massing mutalisks. This prevents a Zerg player to do what's the best - expand early in game.
Most of the Protoss players cannon-up early in game and/or do rushes like dark templar or reaver. To this date i never saw a Protoss player in 1v1 that had expanded before attacking. However, dark templar rush or reaver drop can be very effective if did properly - but properly by expanding and developing in the meantime. Protoss players are also the ones that make the least workers - i don't know why is that, but it's a very big mistake.
Beating 3 Computers
Computers are easy to beat (for me at least :). If you can't beat a random computer without cannoning up then don't get into online games with computers as it will be nothing but a victory or loss for you. Here is one rush for beating 3v1 melee computer game (i'm writing this because allot of people asked for it).
Simply choose a small map (like Blood Bath) and put all races to Zerg, including yourself. Then Build up drones until 6 and morph a spawning pool. Build 3 more drones for 8 control and after the 8th drone has finished send one drone to each computer. As each the drones reach the computer, build creep colonies on their creep, next to hatchery. When the creep colonies finish morphing build sunkens and that's it. By doing this, you'll beat 3 computers under 6 minutes.
The Maynard Transfer
The strategy here is to never stop building workers (overproduce them, having like 3-2 or 4-2). Then when you build your expansion. When it's almost finished (300-400 hit points left) transfer half of your workers to the expansion and you'll have a significant jump in economy. Most of you probably came up with this yourself, as i have, but Maynard was the first player to use it.
Use of the Shift
Shift is one of the most important keys in the game - it cannot be replaced by the mouse. It's wide use is selective firing. Use it when you have units with ranged weapons. Thing here is to attack one unit by pressing the right mouse button, then hold shift and right click the other units to attack. The units will be attacked (and killed) one by one instead of each unit attacking another.
That is the use I mentioned in the Unit Control just above. The second use for the shift is for Protoss and Terran players. You can order a probe to warp in a building, then hold shift and right click on a mineral patch. Your probe will start mining minerals as soon as it opened the warp field. Terran can use shift when repairing units and buildings. Simply hold shift and click on units/buildings that needs to be repaired and the SCV(S)) will start repairing them, one by one.
You can use shift to set way points, what is especially useful when scouting or dropping.