The heart of the problem is that the opportunity cost for using the Defend command is almost always not worth the benefit. When defending, a character typically does nothing for one round, but in exchange takes half damage from all attacks. However, there is one big problem with making this choice: the player has no idea if the defending character will even be attacked that turn. If the player orders a character to Defend, all of the monsters might end up attacking other characters. So, the outcome and net benefit of using the Defend action is unknown. Furthermore, there are alternative strategies that have more reliable outcomes. For example, a wounded character can use a healing item on himself instead of defending. The outcome of that action is more certain: the character will be healed and less likely to die. This is in many ways a superior solution all around, since a defending character is only putting off his own demise, rather than reversing it.
In most RPGs, there is only one scenario where the Defend action is useful: when the player knows that a certain character is going to be hit with an attack capable of killing that character. Typically this happens during specific boss battles where a boss broadcasts that it is going to hit a specific character or the whole team with a super-attack. This is usually a scenario specifically set up by the game designers where using the Defend command is the only viable solution. Scenarios like this include countdown attacks, attacks that require a turn to charge up, or attacks where a boss points out a specific target in some way beforehand.
However, there are a few games where the Defend command is more useful than other games. In these games, the Defend command is paired with an additional effect. For example, the characters in Wild ARMS 3 reload their guns by defending. In Legend of Legaia, the Defend command is called the Spirit command. In addition to decreasing damage by half for the turn, it increases the number of attacks the character can make the next turn. In Legend of Dragoon, the Guard command heals 10% of the characters hitpoints as a bonus. Another good example is Breath of Fire IV, where characters can analyze an enemy's moves while Defending. And yet, while the Defend command is useful in these games, it is usually solely due to the additional effect, rather than the defensive effect. So these solutions don't solve the fundamental problem of the Defend command itself.
In order to make the Defend command useful, particularly as an actual defensive measure, it needs to be reliably useful as a part of a bigger strategy. There are some examples of worthwhile defensive commands in several RPGs. For example, Auron from Final Fantasy X has abilities called Cover and Sentinel. When he uses these commands, he will get in the way of any physical attack aimed at one of his allies. These moves are a great way to protect more vulnerable mage-type characters from big enemies. There is also the various defensive commands from Skies of Arcadia, which I talked about at some length in a previous post. These commands enable a character to "tank" by spending actions to decrease the damage taken by the team as a whole. As such, they are very valuable.
As a whole, the traditional Defend action is useless compared to the advantages of simply healing a wounded character. While there are ways for an RPG game designer to make it useful in particular situations, these solutions generally feel forced. In the long run, it would make for more tactically interesting gameplay to replace the Defend command with dedicated tanking powers.