I'd like to develop this post gradually, as there is a lot of evidence to summarize and refer to.
But here is a start:
"Mood stabilizers" are drugs which are thought to help treat the symptoms of bipolar disorder. It is hoped that these drugs might reduce manic symptoms, prevent recurrence of manic symptoms, while also reducing or preventing symptoms of depression.
The first treatments for manic episodes were sedatives, including barbiturates and antipsychotics.
The first "mood stabilizer" though, was lithium carbonate.
Lithium itself is the third-simplest element in the universe, after hydrogen and helium. It tends to form salts. It is structurally very similar to sodium, which is a salt-forming element essential to most every life function (that is why we find sodium ions abundantly in all body fluids, and in a similar concentration in the ocean; table salt consists of sodium and chlorine atoms which join together as crystals). Yet, lithium is not normally present in the human body, and is much less common in the universe as well, compared to hydrogen, oxygen, carbon, sodium, etc.
The mechanism of lithium's action in the body, when used as a drug, is still poorly understood. Its similarity to sodium is probably essential to its mechanism. I'll add to this commentary later, but for now I will say that the mechanism of lithium salts probably involves multiple actions inside of nerve cells; these actions may modulate cellular activity.
Here are some of the clinical actions of lithium carbonate, when used as a medication in those with bipolar disorder (I will list the actions in order of how clearly proven and substantial the effect is):
1) It reduces symptoms of mania
2) It may reduce the length of a manic episode (note that just because a treatment reduces symptom severity, it may not reduce the duration of the symptoms).
3) It may prevent the recurrence of manic symptoms
4) It may prevent the recurrence of depressive symptoms
5) It may reduce depressive symptoms when they occur
There are other uses for lithium carbonate as well:
1) It can be combined with an antidepressant to improve symptom control in unipolar depression
2) It may help treat specific symptoms such as irritability and rage
3) It helps prevent cluster headaches (a type of severe, recurrent headache)
Lithium is probably most useful in "classic bipolar disorder", in which individuals experience manic episodes with elevated mood (as opposed to irritable or "dysphoric" mood), and in which the mood episodes are not recurring frequently during an average year (i.e. there is no "rapid cycling").
Here are some of the side-effects of lithium:
1) thirst, increased urine production
2) tremor (shaky hands)
4) sedation -- usually it is a much less pronounced type of sedation compared to antipsychotics, benzodiazepines, or other "sleeping pills". But there can be feelings of reduced energy, reduced clarity of thinking, or lethargy
5) toxicity to the kidneys -- this is not common, but needs to be checked for regularly
6) inhibition of thyroid function -- this is not permanent, nor is it harmful to the thyroid gland (in fact, it may "rest" the thyroid gland); but diminished thyroid levels, if present, requires treatment with a thyroid supplement)
7) acne or other skin rashes
8) toxicity in overdose
There have been a few studies questioning the effectiveness of lithium, particularly in terms of its value in preventing recurrent mood episodes. But for many people it does appear to be very effective, both as an acute treatment and as a preventative agent. It probably works much better as an "anti-manic" agent than an "anti-depressant".
There are various forms of lithium, and various dosing regimes. In most cases, it can be dosed simply: once at bedtime. The concentration of lithium in the blood needs to be measured periodically. Levels which are too high can increase the likelihood of toxicity (mind you, excessive levels could usually be assessed on the basis of side-effect complaints); levels which are too low may not be effective.
In my experience, some people may benefit from staying on lithium, but adjusting the dose to a point that is more tolerable for them. It may not necessarily be true that everyone needs to have a full therapeutic concentration of lithium in order for it to work. For some people, the side effects may outweigh the benefit at full doses.
However, it is a frequent situation in an emergency room, or on a mood disorders hospital ward, that people with clear histories of bipolar disorder, stable on medication, end up having a recurrence of severe mania a few weeks or months after tapering or stopping their medication (often lithium). In some of these cases, the manic symptoms may have already been building up, leading the person to discontinue their medication (rather than the other way around). But in many of these cases, it seems to me that the lithium had been protecting them, and that the recurrence of mania happened because of medication discontinuation.
There is also some evidence that sudden lithium discontinuation can provoke increased mood instability. So, while there are no overt withdrawal symptoms from stopping lithium, it should be tapered slowly if possible (I would say over 1-2 months at least).
It should be emphasized that lithium is not a perfect drug, either in terms of side effects or in terms of effectiveness. Many people on full doses of lithium still experience relapses of mania. But it is quite clear, from decades of experience, that lithium can be helpful for many people with bipolar disorder.
(a 2007 Cochrane review of mood stabilizers, showing good evidence for lithium, but also encouraging use of other mood stabilizers--which for some people could be superior to lithium-- such as valproate and atypical antipsychotics)
(a 1994 JAMA article showing the effectiveness of lithium and valproate, compared to placebo, in acute mania)
(a negative study, comparing lithium, valproate, and placebo; published in the major journal Archives of General Psychiatry in 2000--it shows very little difference between lithium, valproate, and placebo treatments with respect to relapses in bipolar patients over a 1-year period; however this study was probably biased in favour of high placebo effects and lower medication treatment effects, for a variety of reasons)
(a randomized, placebo-controlled study from Archives of General Psychiatry in 2000, showing a pronounced effect of lithium in reducing aggression in hospitalized children with conduct disorder)
(a Dutch review article from 2006, and an older article from a U.S. nephrology journal, summarizing the risk of kidney disease associated with lithium; about 15-20% of people taking lithium long-term may experience a decline in kidney function. While this decline is usually mild, I think that an alternative mood stabilizer should be strongly considered if someone is developing signs of reduced kidney function while on lithium).