Stimulants are the most effective treatment in psychiatry by effect size, and the most argued about. This article is what they actually do, what the best current evidence says about how well they work, what they do to blood pressure, and how addictive they really are — plus the one finding that a company selling focus supplements should be most uncomfortable about, which is ours.
Nothing here is dosing advice. We are not doctors or prescribers, the dose figures below are descriptions of what a published dose-response analysis found across trials, not targets for any individual, and nobody should change a prescription on the basis of a blog post.
What a stimulant does mechanically
Dopamine and noradrenaline are released into the synapse and then cleared out of it by transporter proteins — DAT for dopamine, NET for noradrenaline. The prefrontal cortex, which does most of the work we call attention and self-regulation, depends on both.
Methylphenidate (Ritalin, Concerta, Equasym, Medikinet) blocks those transporters. Clearance slows, so dopamine and noradrenaline linger. The analogy to an SSRI is reasonably close — it is a reuptake inhibitor, just on a different pair of neurotransmitters.
Amphetamines (dexamfetamine, lisdexamfetamine/Elvanse/Vyvanse, mixed amphetamine salts) do that and more: they enter the neuron and make the transporters run in reverse, actively pushing dopamine and noradrenaline out into the synapse rather than merely failing to retrieve them. That extra mechanism is why amphetamines tend to produce a larger effect, and also why they carry more abuse potential — pushing transmitter out does not wait for the cell to release any.
Lisdexamfetamine is worth one sentence of its own because the design is clever: it is dexamfetamine bonded to the amino acid lysine, inactive until enzymes in the blood cut the two apart. That makes onset gradual and makes snorting or injecting it largely pointless, which is a deliberate abuse-deterrent feature rather than a marketing claim.
The non-stimulants work differently. Atomoxetine and viloxazine are noradrenaline reuptake inhibitors — NET only, no direct dopamine action. Guanfacine and clonidine are alpha-2 adrenergic agonists, which is a different mechanism again and, as you will see below, has opposite cardiovascular effects.
How well do they work?
The most comprehensive synthesis for adults is a 2025 component network meta-analysis in The Lancet Psychiatry: 113 randomised controlled trials, 14,887 participants, covering drugs, psychological therapies and neurostimulation together.1
At 12 weeks, measured on both self-reported and clinician-reported scales, only two things beat placebo:
- Stimulants — standardised mean difference −0.39 (95% CI −0.52 to −0.26) self-reported, −0.61 (−0.71 to −0.51) clinician-reported.
- Atomoxetine — −0.38 (−0.56 to −0.21) self-reported, −0.51 (−0.64 to −0.37) clinician-reported.
CBT, cognitive remediation, mindfulness, psychoeducation and transcranial direct current stimulation all beat placebo on clinician-reported measures but not on self-reported ones. That split is important and awkward: it means the people receiving those treatments did not consistently report feeling different, even where the assessing clinician scored them as improved. Atomoxetine and guanfacine were also less acceptable than placebo — more people stopped taking them.
Two findings from that paper deserve more attention than they get:
Medications were not efficacious on quality of life. They reduce core symptom scores. The review found no demonstrated benefit on the broader outcome most people actually care about. That is not the same as saying they don't help anyone's life — it is saying the trials have not shown it, which is a gap in the evidence rather than a verdict.
Longer-term evidence is underinvestigated. Nearly all of this is 12 weeks or less, for a condition treated across decades. The authors say so plainly, and the certainty ratings across the whole network ranged from very low to moderate.
Dose: what the newest analysis found, and why it cuts both ways
A 2026 dose-effect network meta-analysis in The Lancet Psychiatry — 113 RCTs, 14,138 children and adolescents and 11,016 adults — modelled efficacy and tolerability as curves against dose rather than comparing fixed doses.2 In children and adolescents, median efficacy rose up to roughly 45 mg/day for methylphenidate, 25 mg/day for amphetamines and 4 mg/day for guanfacine, with no evidence of further benefit above those points — though the authors note the credible intervals at higher doses were wide. In adults, amphetamines plateaued above about 50 mg/day, while methylphenidate showed no clear plateau, possibly because the data thin out. Discontinuation from side effects rose with dose for amphetamines (above 25 mg/day in children, 50 mg/day in adults) and for methylphenidate above 50 mg/day in adults. Atomoxetine and modafinil showed no dose-effect pattern at all in fixed-dose studies.
The authors' framing is the valuable part, and it points in two directions at once: the findings challenge both therapeutic inertia — accepting a poor response without trying a higher dose — and uncritical escalation beyond licensed limits. Those are opposite errors and the same paper argues against both. If you are on a stimulant and it is not working, "the dose might be too low" and "more is not automatically better" are both legitimate things to raise with a prescriber. Neither is a conclusion you can reach from a webpage.
Blood pressure and pulse: the finding that should change a common assumption
A 2025 network meta-analysis in The Lancet Psychiatry pooled 102 RCTs — 13,315 children and adolescents, 9,387 adults, median follow-up 7 weeks — on what ADHD medications do to haemodynamics.3
Amphetamines, atomoxetine, lisdexamfetamine, methylphenidate and viloxazine all raised blood pressure or pulse or both. In children and adolescents, the mean increase against placebo ran from +1.07 mmHg systolic (atomoxetine) to +1.81 (methylphenidate); diastolic from +1.93 (amphetamines) to +2.42 (methylphenidate); and pulse from +2.79 bpm (viloxazine) to +5.58 bpm (atomoxetine). In adults, systolic +1.66 to +2.3, diastolic +1.60 to +3.07, pulse +4.37 to +5.8.
Here is the part that matters. Amphetamines, lisdexamfetamine and methylphenidate were not associated with larger increases than atomoxetine or viloxazine, in either age group. The largest pulse increase in children in this analysis belonged to atomoxetine — a non-stimulant. If you have switched, or are considering switching, from a stimulant to a non-stimulant believing it is gentler on the heart, this analysis does not support that. The authors' own recommendation is explicit: monitor blood pressure and pulse in anyone on any ADHD medication, not stimulants only.
Guanfacine is the exception, and it moves the other way — reducing systolic pressure by a mean 2.83 mmHg in children and 10.1 mmHg in adults versus placebo, which is a reminder that it is a blood-pressure drug by origin. Only four trials informed medium-term effects and none informed long-term ones.
To keep the numbers in proportion: an average rise of one to three mmHg across a population is small, and for most people it is clinically unremarkable. It is also an average, which means some individuals sit well above it — which is the entire argument for measuring rather than assuming.
How addictive are they, actually?
A 2026 systematic review in Pharmaceuticals synthesised 35 studies on methylphenidate's addictive potential — behavioural, clinical, neuroimaging and preclinical.4 Its conclusions are more specific than either side of the public argument:
- Methylphenidate does produce measurable reinforcing effects and subjective drug liking, closely tied to craving — and these depend on dose, pharmacokinetics and route of administration. Rapid dopamine rises, especially from non-oral or high-dose use, carry higher abuse potential.
- Appropriately monitored therapeutic use was associated with a low observed risk of substance use disorders.
- Misuse was more common among people without ADHD, and was typically motivated by wanting cognitive enhancement rather than by euphoria.
- Pharmacodynamic tolerance appeared in a subset of patients. A classical withdrawal syndrome was not consistently observed — what occurred was generally the return of the original symptoms.
The practical shape of that: the risk is concentrated in high doses, non-oral routes and non-prescribed use, which is why extended-release formulations and monitoring are the standard risk-management answer. "Stimulants are addictive" and "prescribed stimulant treatment is low-risk" are both supported by this evidence, because they are statements about different situations.
The finding our own industry should sit with
Since the misuse evidence says non-medical stimulant use is mostly about cognitive enhancement rather than getting high, the obvious question is whether it delivers.
A 2026 study in Experimental and Clinical Psychopharmacology gave 183 college students a placebo while telling them it was 10 mg of Adderall, and varied whether they got to choose to take it.5 Almost everyone offered the choice took it. The results:
- People who believed they had taken a stimulant reported enhanced mood and drug effects on several measures — medium-to-large placebo effects.
- On objective cognitive tasks there were essentially no placebo effects at all, with one exception on a single reaction-time index. Choosing to take it made no difference either.
Believing you have taken something that sharpens your focus reliably changes how sharp you feel. It does not change how well you actually perform. That is the cleanest demonstration we have seen of the gap between felt and measured benefit — and it is precisely the gap the entire nootropics category, ours included, operates inside.
We sell Cognitive Support. It contains L-tyrosine, which is the amino acid your body converts into dopamine and noradrenaline — the same two neurotransmitters a stimulant acts on. That shared biochemistry is real, and it is also where honest companies and dishonest ones part company, because a precursor is not a reuptake inhibitor. Supplying more raw material to a pathway is not the same intervention as changing how fast the pathway clears. The human evidence on tyrosine is narrower than that mechanism story implies, and we went through what it does and does not show in L-tyrosine for focus and stress.
So, plainly: Cognitive Support is not an alternative to ADHD medication, and we are not going to position it as one. The effect sizes in section three above are for prescription drugs in diagnosed ADHD, and nothing in our range has been tested against them or comes close to that evidence base. If you suspect you have ADHD, the route is assessment and a prescriber, and our pieces on what the UK wait actually looks like and the Irish equivalent are honest about how long that takes. Eight of Cognitive Support's ten active ingredients sit in a 255mg blend whose individual amounts we cannot publish, which we spell out in inside Cognitive Support — and if you want the broader argument about what an ingredient has to clear before we will put it on a label, it is in why we don't sell brain blends.
If you take both: caffeine is the live question
The combination people actually ask us about is a prescribed stimulant plus caffeine, whether from coffee or a focus supplement. We looked at this properly in Cognitive Support and ADHD medication and the honest finding bears repeating: no controlled study of caffeine alongside a prescribed stimulant exists. That is not the same as a study finding it safe. Both raise heart rate and blood pressure, both can disrupt sleep, and the cardiovascular numbers in section four are the baseline before you add anything. If you are stacking them, that is worth mentioning at the appointment where someone takes your blood pressure.
Sleep is the other place these collide, and it runs in both directions — ADHD and sleep problems overlap heavily on their own, before medication enters it. That overlap is its own article.
The short version
- Methylphenidate blocks dopamine and noradrenaline reuptake. Amphetamines do that and also reverse the transporters, which is why they hit harder and carry more misuse potential.
- In adults, stimulants and atomoxetine are the only treatments that beat placebo on both self- and clinician-rated symptoms. Stimulant SMD: −0.39 self-reported, −0.61 clinician-reported.
- No demonstrated benefit on quality of life, and almost no long-term evidence. Both are evidence gaps, not disproof.
- Efficacy stops climbing around 45 mg/day methylphenidate and 25 mg/day amphetamines in children. The same analysis argues against under-dosing and against escalating past licensed limits.
- All the main drugs raise blood pressure and pulse by small amounts — and stimulants are no worse than atomoxetine or viloxazine on this. Guanfacine lowers them. Monitor on any of them.
- Monitored therapeutic use carries low addiction risk. Misuse is commoner in people without ADHD and is usually about performance, not euphoria.
- Believing you took a stimulant makes you feel sharper without making you measurably sharper. Every focus supplement on the market, ours included, has to be read against that.
The rest of this series
This is one of four explainers on the medication classes that come up most often alongside what we sell. Each one is the mechanism, the best current evidence, and the places where that evidence is weaker or stranger than the standard account — including where it counts against us.
- What SSRIs actually do — the serotonin transporter, why the "chemical imbalance" account is disputed in print, and why published withdrawal estimates range from about 15% to 55%.
- What benzodiazepines actually do — GABAA amplification, why Z-drugs are not a separate category, and why stopping badly is the genuinely dangerous part.
- What antipsychotics actually do — four dopamine pathways and the side effects that follow, tardive dyskinesia rates, and the one prescribed off-label as a sleeping tablet.
This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease. We are not doctors, pharmacists or prescribers. The dose figures above describe findings from a published dose-response analysis across clinical trials; they are not recommendations, not targets, and not applicable to any individual. Nothing here is a reason to start, stop, change or skip a prescribed medication — including changing a dose. ADHD medication should be started, adjusted and monitored by a qualified prescriber, and taking a stimulant that was not prescribed to you carries risks this article does not cover. If you take any prescription medication, speak to your doctor or pharmacist before taking any supplement.
References
- Ostinelli EG, Schulze M, Zangani C, Farhat LC, Tomlinson A, Del Giovane C, et al. "Comparative efficacy and acceptability of pharmacological, psychological, and neurostimulatory interventions for ADHD in adults: a systematic review and component network meta-analysis." The Lancet Psychiatry, 2025, 12(1):32–43. doi:10.1016/S2215-0366(24)00360-2
- Nourredine M, Jurek L, Hamza T, Cipriani A, Subtil F, Parlatini V, et al. "Pharmacological interventions for ADHD: a systematic review and dose-effect network meta-analysis." The Lancet Psychiatry, 2026, 13:485–495. doi:10.1016/S2215-0366(26)00091-X
- Farhat LC, Lannes A, Del Giovane C, Parlatini V, Garcia-Argibay M, Ostinelli EG, et al. "Comparative cardiovascular safety of medications for attention-deficit hyperactivity disorder in children, adolescents, and adults: a systematic review and network meta-analysis." The Lancet Psychiatry, 2025, 12(5):355–365. doi:10.1016/S2215-0366(25)00062-8
- Bieś R, Chabrzyk A, Zborowska K, Krysta K, Krzystanek M. "Addictive Potential of Methylphenidate: A Systematic Review of Human and Preclinical Behavioral, Clinical and Neurobiological Evidence." Pharmaceuticals, 2026, 19(9):1409. doi:10.3390/ph19091409
- Looby A, Berry KA, Petrey AM, Winterlind EL. "To choose or not to choose: Examining the influence of choice on elicitation of subjective and objective prescription stimulant placebo effects in the laboratory." Experimental and Clinical Psychopharmacology, 2026, 34(1):44–51. doi:10.1037/pha0000824