What actually helps anxiety, ranked honestly

Anxiety attracts an unusual density of advice, most of it confident and some of it useless. It is worth separating what has been tested properly from what has been repeated often.

The strongest evidence

Cognitive behavioural therapy. The most robustly supported treatment for most anxiety disorders, with effects that persist after treatment ends better than medication does. The active ingredient in most anxiety CBT is exposure — systematically approaching what you avoid — rather than thought-challenging, which is the part people expect. Twelve to sixteen sessions is a typical course. Digital CBT with therapist support has decent evidence and is more accessible.

SSRIs and SNRIs. First-line medication. They take four to six weeks for full effect and often make anxiety slightly worse in the first fortnight, which is the single most common reason people stop before they work. Starting low and going slow reduces that. They are not sedatives and they are not addictive in the dependence sense, though stopping abruptly produces discontinuation symptoms and should be tapered.

Exercise. Genuinely effective, with effect sizes in trials that compare respectably to medication for mild to moderate anxiety. The evidence supports both aerobic exercise and resistance training. Frequency and consistency matter more than intensity, and the effect appears within weeks. Strength training does more for women than any other exercise covers the wider case.

Sleep. Not a treatment for anxiety exactly, but sleep deprivation reliably raises next-day anxiety in experimental studies, and insomnia and anxiety maintain each other. Treating the insomnia often improves the anxiety, and CBT-I is the tool for that — the first-line insomnia treatment most people have never heard of.

Moderate evidence, worth trying

Mindfulness-based stress reduction. Real effects in trials, generally smaller than CBT, and dependent on actually doing the practice. Apps have thinner evidence than the structured eight-week programmes they are modelled on.

Reducing alcohol. The rebound mechanism means alcohol reliably increases anxiety the following day, and people using it to manage anxiety are usually in a self-sustaining loop. Alcohol, sleep and next-day anxiety explains why.

Reducing caffeine. Caffeine is anxiogenic at sufficient doses, and sensitivity varies enormously between people for genetic reasons. If you are anxious and drinking four coffees, this is a cheap experiment. Caffeine: your half-life, not the average one.

Treating the physical causes. Hyperthyroidism, iron deficiency and sleep apnoea all present as anxiety. A basic blood panel and an honest look at sleep are worth doing before concluding the problem is purely psychological — low ferritin and low mood and fatigue that sleep doesn't fix cover it.

Weak or contested

Most supplements. Magnesium, ashwagandha, valerian, L-theanine and others have small, low-quality or inconsistent trial evidence. Some may do something modest; none has support comparable to the interventions above. Magnesium is cheap, safe for most people and worth a try if you want one, but it is not a treatment.

CBD. The clinical evidence in anxiety disorders remains thin despite the volume of marketing. Product quality and dosing are wildly inconsistent.

Breathing exercises. Slow breathing does acutely reduce physiological arousal — that part is real. As a standalone treatment for an anxiety disorder, it is not sufficient, though it is a reasonable component.

What actively makes it worse

Avoidance. The single most important maintaining factor. Every avoided situation confirms the prediction that it was dangerous, and the anxiety grows to fill the space. This is why exposure is the core of effective treatment.

Reassurance seeking, including repeated symptom-checking and asking the same person the same question. It functions identically to avoidance.

Benzodiazepines beyond short-term use. Effective acutely, but tolerance develops, they interfere with the learning that exposure therapy depends on, and withdrawal is difficult. Appropriate for short, defined situations; not a long-term answer.

The part specific to women

Anxiety is roughly twice as common in women, and one contributor is frequently missed: cyclical anxiety. For a meaningful proportion of people, anxiety is not constant but concentrated in the late luteal phase, driven by neurosteroid fluctuation — anxiety before your period.

The treatment implication is significant. Cyclical anxiety may respond to ovulation suppression or to luteal-phase-only SSRI dosing, neither of which is offered to someone whose anxiety is assumed to be continuous. And continuous anxiety that worsens premenstrually needs continuous treatment.

You cannot tell these apart from memory. It requires daily ratings across at least two cycles — including the days you feel fine, because the pattern is defined by the contrast.

The uncomfortable practical point

Almost everything above takes weeks to show an effect, and human memory for mood is bad enough that people frequently cannot tell whether something worked. Plenty of people stop an SSRI at week three, or abandon exercise, or conclude therapy did nothing, on the basis of a recollection that is simply wrong.

A daily record fixes that. It is unglamorous and it is the difference between running an experiment and having an impression.

Naked is built for that: rating anxiety daily against sleep, cycle, alcohol, exercise and whatever else you are trying, so that in three months you can see what actually moved rather than argue about it with yourself.