Psychiatric medication half-life quick reference

Why this matters clinically

Three practical questions rest on half-life. First, how fast can I taper without causing withdrawal? Roughly, half-life under 12 hours means slow and careful. Second, how long until steady state? About five half-lives, which for most SSRIs is a week or so but for fluoxetine is a month and for cariprazine is a couple of months. Third, when I switch or add a drug with interaction potential, how long is the previous drug still on board? Fluoxetine plus norfluoxetine can suppress 2D6 for four to five weeks after the last dose.

SSRIs

SNRIs

Atypical antidepressants

TCAs

Wide individual variability makes therapeutic drug monitoring worth doing. Levels drawn 10 to 14 hours after the last dose are the usual timing.

MAOIs

Half-life is misleading here because the irreversible MAOIs bind their enzyme and stop working when the enzyme is regenerated, not when the drug clears. Plan on 14 days of monoamine oxidase recovery after stopping phenelzine, tranylcypromine, or isocarboxazid before starting a serotonergic drug. After stopping fluoxetine, wait five weeks before starting an MAOI. After stopping other SSRIs and most SNRIs, wait two weeks. Selegiline transdermal at 6 mg/24 h doesn't require the dietary tyramine restrictions, but the washout rules still apply.

Antipsychotics

Long-acting injectable antipsychotics

For LAIs, the number that matters is apparent half-life (how long the depot releases drug into the bloodstream), which is much longer than the parent drug's own half-life. Steady state typically requires four to five apparent half-lives.

Practical taper timing: after the last LAI dose, plan on months, not weeks, before the drug is fully out. This matters when switching, when a patient becomes pregnant, or when you need to rule out drug-induced parkinsonism.

Benzodiazepines

Half-life picks the clinical use here. Short-acting agents are more addictive and produce sharper withdrawal. Long-acting agents smooth things out but accumulate, especially in the elderly.

Z-drugs

Mood stabilizers

Stimulants

Effective duration matters more than parent half-life for stimulants because of formulation choices and metabolite activity.

What half-life actually tells you about tapering

The relationship is unfortunately simple. Short half-life means the drug leaves fast between doses, which means withdrawal shows up between doses even at steady dosing (interdose withdrawal, seen with alprazolam and IR venlafaxine) and shows up dramatically when the drug stops. Long half-life means the drug tapers itself. Fluoxetine's month-long tail is why fluoxetine discontinuation syndrome is rare, and why "cross-tapering" from paroxetine to fluoxetine and then stopping fluoxetine is a legitimate technique for a stuck paroxetine taper.

Rough rules:

• Half-life under 12 h: taper slow (10 to 25 percent per step), pause between steps (2 to 4 weeks), consider liquid or bead-counting at the tail end.
• Half-life 12 to 48 h: standard taper works for most patients (25 percent per step every 2 to 4 weeks).
• Half-life over 48 h: taper is often unnecessary; step down or stop, since the drug tapers itself.

Fluoxetine and aripiprazole are the two most forgiving common drugs in this respect. Paroxetine and venlafaxine IR are the two least forgiving.

Time to steady state

Steady state is reached at about five half-lives. That's a rough rule, not a law, and it assumes linear kinetics (which don't quite apply for paroxetine, fluoxetine, or lithium in older patients).

Rough guide:

• Most SSRIs and SNRIs: 1 week (fluoxetine is the exception; norfluoxetine keeps rising for a month).
• Most oral antipsychotics: 4 to 7 days.
• Aripiprazole: 2 weeks (the parent) to 3 weeks (including active metabolite).
• Cariprazine: 6 to 8 weeks. This one gets missed. Do not conclude non-response at week 2.
• Lithium: 5 to 7 days.
• Carbamazepine: fluid target because of autoinduction. Recheck.
• Lamotrigine: about a week at each step. Titration is safety-driven, not level-driven.
• LAIs: months for most, up to about six months for paliperidone palmitate monthly without loading.

The clinical implication is that whenever you change a dose, you're deciding what to do about a version of the patient's drug exposure that won't fully arrive for days to weeks. That's part of why psychiatric prescribing rewards patience.

Common questions

Why does fluoxetine forgive a missed dose but paroxetine doesn't? Norfluoxetine, the active metabolite of fluoxetine, has a half-life of 7 to 15 days. That means a single missed dose barely moves the total exposure. Paroxetine has no active metabolite and a parent half-life around 21 hours, which shortens further at lower doses because of nonlinear kinetics. Miss a dose or two and the drug level plummets. That's why the "half-life crash" from paroxetine (dizziness, brain zaps, flu-like feeling) shows up within a day or two while fluoxetine discontinuation is often invisible.

How long after stopping fluoxetine before I can start an MAOI? Five weeks. The reason is norfluoxetine, which takes that long to clear. Starting an MAOI while any meaningful fluoxetine or norfluoxetine remains risks serotonin syndrome. For sertraline, paroxetine, citalopram, escitalopram, and venlafaxine, two weeks is the standard washout. Going the other way (stopping an MAOI, starting an SSRI or SNRI) is 14 days for phenelzine and tranylcypromine, because MAO enzyme has to regenerate.

Why does clozapine build up on some patients but not others? CYP1A2 activity varies substantially between people. Smoking induces 1A2 and drops clozapine levels; women and non-smokers have higher levels on the same dose. Add fluvoxamine or ciprofloxacin (both strong 1A2 inhibitors) and levels rise sharply. Age, obesity, and inflammation (which suppresses 1A2 in acute illness) all shift things too. That's why clozapine is one of the few psychiatric drugs where levels are essential rather than optional. Draw a trough, adjust to the 350 to 600 ng/mL range, and repeat whenever anything changes (smoking status, a new antibiotic, an infection).

When can I judge if aripiprazole is really working? Not before four weeks, and often longer. Aripiprazole and dehydroaripiprazole have half-lives of 75 and 94 hours, so steady state takes about three weeks. Behavior takes another week or two to catch up. Patients and family who decide at day 10 that "it's not doing anything" are looking at a drug that hasn't reached steady state yet. For cariprazine, extend that to 6 to 8 weeks because the active metabolites keep rising for weeks.

How long until an LAI reaches steady state? Depends on the LAI and whether you loaded. Paliperidone palmitate monthly (Sustenna) with the standard loading dose regimen (day 1, day 8) is at steady state within a couple of months. Without loading, closer to six months. Aripiprazole Maintena needs about four months and requires two weeks of oral overlap on the first injection. Haloperidol decanoate reaches steady state at about three months. In practical terms, do not judge whether an LAI is working before the drug has actually gotten there. A patient who "isn't responding" to their third Maintena injection is not yet at steady state.

Sources

1. FDA Prescribing Information for each medication listed. Available via DailyMed (National Library of Medicine).
2. Stahl SM. Stahl's Essential Psychopharmacology, 7th edition. Cambridge University Press.
3. Bazire S. Psychotropic Drug Directory. Current edition.
4. Cochrane reviews on antidepressant discontinuation syndromes and taper strategies.
5. Horowitz MA, Taylor D. Tapering of SSRI treatment to mitigate withdrawal symptoms. Lancet Psychiatry 2019.
6. Correll CU. Long-acting injectable antipsychotics: focus on olanzapine pamoate, paliperidone palmitate, aripiprazole. Multiple review articles.
7. American Psychiatric Association. Practice Guideline for the Treatment of Patients with Schizophrenia, third edition.
8. Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines for CYP2D6 and CYP2C19 substrates.

Reviewed against current guidelines as of June 8, 2026. This is not medical advice.