It depends on your requirements. With databases and email application servers some people do nothing but fulls. Some do a full/inc or full/diff strategy. Replaying transaction logs in a database or email application can be particularly time consuming come restore time if you have incrementals (backups of the transaction logs). In general, file system backups are usually a weekly full, daily inc strategy.
Fulls consume more media but restore faster.
Incrementals consume less media and backup faster, but restores take longer because you must first restore the full, then all incrementals to whatever point in time you're trying to restore.
Having some knowledge of your environment also helps. If you know the given server is fairly static and not much changes, you'll know your incrementals will be particularly small and restore times probably won't be much different then doing a full restore. On the other hand if you have a lot of data changing, it's possible your incrementals will be almost the size of your fulls and then it gets ugly.
Inexperienced people have a tendancy of focusing too much on backup performance and media consumption and don't put enough weight on restore time. Imagine how your policies and decisions (equipment or otherwise) will play out if you have a minor, medium, or major disaster and is that acceptable? Isn't that truly what you're going through all this for?