Most people taking magnesium for sleep or brain health are taking a form that cannot reach the brain in meaningful concentrations. Magnesium glycinate, the most commonly recommended form in integrative medicine, is absorbed in the small intestine via passive diffusion. It raises peripheral serum magnesium. It does not significantly elevate brain magnesium. Neither does citrate, oxide, or malate. The blood brain barrier has no efficient passive transport mechanism for most magnesium anions.

This is not a minor bioavailability difference. Brain magnesium regulates NMDA receptor activity, synaptic density in the hippocampus, and the GABAergic signaling that initiates cortical quieting before sleep. Peripheral magnesium does not substitute for central magnesium. You can have a completely normal serum magnesium reading and still have functionally depleted brain magnesium. Serum magnesium reflects what is circulating in blood, not what is inside neurons.

Magnesium L-threonate (the compound is L-threonic acid magnesium salt, sold under the patented name Magtein) is the one form that crosses the blood brain barrier efficiently, via a mechanism that no other magnesium salt replicates. Rhonda Patrick has done a dedicated FoundMyFitness episode specifically on the gap between animal and human evidence for this compound. That gap has narrowed considerably since 2025. Here is what the research currently shows.

The blood brain barrier actively restricts central nervous system entry. Magnesium crosses in very limited quantities under baseline conditions. The brain maintains its own dedicated regulation of intraneuronal magnesium, largely independent of serum levels.

Threonate, the anion that pairs with magnesium in L-threonate, is naturally present in cerebrospinal fluid. When you take magnesium L-threonate orally, plasma threonate rises and threonate is transported across the blood brain barrier via glucose transporters (GLUTs), the same active transport proteins that move glucose into neurons. Once inside neurons, elevated threonate directly increases intraneuronal magnesium concentration. Sun and colleagues (2016, Neuropharmacology, 48 citations) demonstrated this in hippocampal neuron cultures: threonate treatment elevated intracellular Mg2+, and this effect was entirely blocked by GLUT inhibitors. Other magnesium anions (citrate, glycinate, oxide, taurate) do not use this pathway and do not produce the same intraneuronal elevation. The threonate ligand is not interchangeable.

The functional consequence of elevated intraneuronal Mg2+ is well established. Magnesium acts as a voltage dependent blocker of NMDA receptors at resting membrane potential. When intraneuronal magnesium is adequate, NMDA receptors open only under conditions of strong concurrent synaptic activity. This gating mechanism sharpens signal to noise in synaptic transmission. Only synapses with strong concurrent activity pass the threshold for long term potentiation. When brain magnesium drops, this gate loosens. NMDA channels open more readily, excitotoxic calcium influx increases, and the synaptic specificity that underlies learning and memory degrades. Elevating brain magnesium via threonate restores the gate.

Sun et al. also found that threonate upregulates NR2B containing NMDA receptors specifically, the subtype most associated with synaptic plasticity and long term potentiation, and increases functional synapse density in hippocampal neuron cultures. Higher synapse density is the structural correlate of cognitive reserve. More active synaptic connections means more redundancy and more processing capacity in the circuits that support memory formation and retrieval.

Li and colleagues (2014, Molecular Brain, 134 citations) extended this to an Alzheimer's disease mouse model. MgT treatment reduced amyloid beta plaque burden, prevented synapse loss, and reversed cognitive deficits. These benefits were maintained even when treatment began at the end stage of the disease progression. The mechanism was NMDA receptor preservation: MgT prevented calcineurin overactivation (the phosphatase that degrades synaptic NMDA receptors under chronic low magnesium conditions) and stabilized BACE1 expression. This is animal data and cannot be extrapolated directly to humans, but the mechanistic clarity it provides is relevant context for the human trials.

The strongest human evidence is Lopresti and colleagues (2026, Frontiers in Nutrition), a 6-week, double blind, placebo controlled trial in 100 adults aged 18 to 45, all with self reported dissatisfied sleep. Dose: 2g per day of Magtein. Primary cognitive endpoint: the NIH Cognitive Toolbox, a validated computerized battery used across NIH research programs. Results: Magtein produced significant improvement in NIH Total Cognition Composite (p=0.043), with larger treatment effects on working memory and episodic memory specifically. Estimated brain cognitive age was reduced by 7.5 years compared to placebo. Reaction time improved (p=0.031). Objective heart rate variability measured via Oura Ring increased (p=0.036), indicating improved autonomic regulation and reduced physiological stress load. The compound was well tolerated with no significant adverse events.

The sleep findings from this trial were mixed and worth stating precisely. Self reported sleep related impairment improved (p=0.043). A subset with more severe sleep problems showed significant improvements in sleep disturbances (p=0.031). But Oura Ring objective sleep staging data showed no significant group differences in sleep duration, deep sleep percentage, or REM sleep percentage in the full sample. The sleep effect in the Lopresti 2026 data operates through autonomic regulation (lower resting heart rate, higher HRV) rather than directly measurable changes in sleep architecture for the average participant.

Hausenblas and colleagues (2024, Sleep Medicine: X, 12 citations) ran a trial specifically designed around sleep, with 80 adults aged 35 to 55 with self assessed sleep problems, using 1g per day MgT for 21 days. Oura Ring data showed significant improvements versus placebo in deep sleep score, REM sleep score, light sleep time, activity score, and readiness parameters. Subjective measures confirmed improved behavior upon awakening, energy, daytime productivity, mood, and mental alertness. The divergence from Lopresti may reflect population selection: the Hausenblas cohort was older (35 to 55), specifically selected for sleep problems, and in an age range where genuine sleep architecture deficits tied to declining GABAergic tone are more prevalent. The compound likely produces more measurable sleep staging changes when baseline sleep architecture is meaningfully compromised.

Zhang and colleagues (2022, Nutrients, 28 citations) used a Magtein plus phosphatidylserine combination formula in 109 healthy Chinese adults aged 18 to 65, at 2g per day for 30 days. All five subcategories of the Clinical Memory Test improved significantly versus placebo. The older participants showed greater improvement than younger participants, the expected direction if the primary mechanism involves restoring depleted brain magnesium rather than providing surplus in people whose levels are already adequate.

Rhonda Patrick has been publicly cautious about magnesium L-threonate's brain benefits in humans, and her position is worth addressing directly rather than ignoring. In a dedicated FoundMyFitness episode on MgLT animal versus human evidence, she noted one controlled trial in 44 participants taking 1,500 to 2,000mg per day for 12 weeks: plasma Mg rose marginally, red blood cell Mg showed no change, urinary magnesium was elevated substantially (suggesting the majority was excreted), and cognitive performance did not improve compared to placebo. She concluded the brain benefits seen in animal models had not been established in humans as of that episode, and that MgLT should not be counted toward daily RDA needs.

That trial is a legitimate data point. The Lopresti 2026 study, a larger sample with the validated NIH cognitive composite as primary endpoint using Magtein specifically, is the strongest counter-evidence now available. What may explain the earlier null result: the 44-person trial likely included healthy younger adults without meaningful magnesium deficiency, where a ceiling effect prevents detectable improvement. The signal from Lopresti and Zhang both runs in the same direction. Older participants and those with sleep or cognitive deficits show greater effects, consistent with a restoration mechanism rather than an enhancement mechanism.

Patrick's core point about RDA remains valid. The 2g per day Magtein dose delivers approximately 144mg of elemental magnesium. This should not be counted toward your daily magnesium recommended intake. If you are also concerned about general magnesium deficiency, which affects approximately 50% of adults in developed countries, supplement that need separately with magnesium glycinate or malate. MgLT is a targeted brain Mg intervention, not a general mineral replacement.

The meta-analysis from Chen and colleagues (2024, Advances in Nutrition, 24 citations) adds a critical constraint: serum magnesium shows a U shaped association with all cause dementia risk (Pquadratic=0.003). The optimal serum Mg is approximately 0.85 mmol/L. Both low serum Mg (below 0.75 mmol/L) and high serum Mg (above 0.95 mmol/L) are associated with elevated dementia risk, with hazard ratios of 1.43 and 1.30 respectively versus the optimal range. More magnesium is not unconditionally better. The goal is optimal brain magnesium, not maximum magnesium.

Magnesium's sleep initiation effect operates through two converging pathways, and understanding both clarifies when you should expect it to work and when you should not.

First: Mg2+ regulates NMDA receptor tone throughout the cortex. When brain magnesium is adequate, NMDA mediated cortical excitation is properly gated, reducing the background excitatory drive that competes with sleep onset. When brain Mg is low, cortical excitability increases and the GABAergic system must work against a higher baseline of excitation. This is a mechanistic explanation for why individuals with low brain magnesium commonly report difficulty quieting mental chatter at bedtime. The inhibitory system is working against a higher baseline of NMDA mediated arousal.

Second: magnesium directly modulates GABA A receptor function. Mg2+ acts as a positive allosteric modulator of GABA A receptors in specific brain regions including the forebrain areas that regulate wakefulness. Enhanced GABA A activity in these regions reduces the drive to maintain wakefulness. The effect is dose dependent and modest, which is why the sleep benefit of magnesium L-threonate appears most reliably in individuals with documented sleep architecture problems rather than in people who already sleep well, exactly the pattern seen across the Lopresti and Hausenblas trials.

The HRV increase from the Lopresti 2026 data (p=0.036) supports this model. Higher HRV reflects reduced sympathetic tone and greater parasympathetic activity, the autonomic signature of a nervous system that has transitioned out of threat detection mode. Whether you are using Magtein primarily for cognitive aging or sleep, improved autonomic regulation appears to be the shared downstream mechanism.

Magnesium L-Threonate (as Magtein, 2g per day). Look for "Magnesium L-Threonate as Magtein" explicitly on the label. Magtein is the patented, trademarked form used in every human RCT showing cognitive and sleep results. Generic magnesium L-threonate products may or may not use the same material at the same purity. Only products listing Magtein have been validated in controlled trials. Most products providing 2g per day come in servings of three capsules. Confirm third party identity testing. Do not count this dose toward your general magnesium RDA. Treat it as a targeted brain Mg intervention and supplement general magnesium needs separately with glycinate or malate if deficiency is a concern.

I have reviewed the available options against the clinical evidence and the most reliably standardized forms are available through my Fullscript dispensary at a discount below retail. Magtein is included alongside the full cognitive and longevity stack I use and recommend.

If this issue was useful, forward it to one colleague who tracks their health seriously. That is the most direct way to help this newsletter grow.

Have a question about the research or a specific stack you want me to look at? Reply to this email. I read every response.