Both first-line drugs earned their Level A grade from randomized, double-blind, placebo-controlled evidence, and both grades are appropriate on the trials that exist. For propranolol, the beta-adrenergic antagonist, controlled trials established a significant effect versus placebo; in one classic double-blind crossover study, only propranolol — not phenobarbital — was significantly more effective than placebo (Baruzzi et al., Neurology, 1983). For primidone, the barbiturate-derived anticonvulsant, a double-blind, placebo-controlled trial confirmed efficacy in hand and head tremor (Findley et al., J Neurol Neurosurg Psychiatry, 1985), and further controlled work showed reduction across a wide dose range (Koller & Royse, Neurology, 1986). The American Academy of Neurology's evidence-based review synthesized this literature into the Level A rating both agents still carry (Zesiewicz et al., Neurology, 2011).

The important caveat sits inside that pedigree: these are the pivotal trials, and they are old and small. The Baruzzi and Findley studies enrolled on the order of a dozen patients each. That does not invalidate the effect — the signal has been replicated — but it means the precision, subgroup resolution, and long-horizon data that a modern regulatory dossier would carry are simply absent here.

The two agents reduce tremor by different, incompletely understood routes, and that difference partly explains their distinct tolerability ceilings. Propranolol is a non-selective beta-adrenergic antagonist; its anti-tremor effect is generally attributed to peripheral beta-2 receptor blockade at the muscle spindle, with a probable central contribution, which is consistent with the observation that non-selective beta-blockers outperform beta-1-selective ones for tremor (Schneider & Deuschl, Neurotherapeutics, 2014). That peripheral mechanism is also why its dose-limiting effects are cardiorespiratory rather than sedative. Primidone is metabolized to phenobarbital and phenylethylmalonamide (PEMA), and its action is thought to involve GABA-A receptor modulation, though the parent compound itself appears to carry much of the acute anti-tremor effect — benefit is often seen before phenobarbital accumulates (Hopfner & Deuschl, Neurotherapeutics, 2020). Neither mechanism targets the central oscillatory network (the cerebello-thalamo-cortical circuit) thought to generate the tremor directly, which is a mechanistic reason to expect attenuation rather than abolition — and a reason first-line pharmacotherapy has a ceiling that procedural options such as thalamic DBS or focused ultrasound thalamotomy can exceed in refractory disease.

The honest summary is meaningful but partial. Quantitatively, short-term studies place the reduction in the moderate range rather than near-abolition. A study of a single primidone dose found tremor reduced by 54% to 69% in eight of eleven patients tested (Seyfert & Honé, J Neurol, 1988); propranolol's controlled-trial effect is significant but similarly incomplete (Baruzzi et al., Neurology, 1983). Complete suppression is the exception, not the expectation. A clinically useful framing for patients — and a realistic one to set at initiation — is functional improvement (steadier writing, feeding, and daily tasks) rather than a still hand.

One counterintuitive and practically useful finding: with primidone, low doses were as effective as high doses in controlled testing, with no correlation between therapeutic response and serum levels (Koller & Royse, Neurology, 1986). The clinical implication is that pushing the dose is not reliably where additional benefit is found, and that the tolerability cost of higher doses often buys little.

The single most clinically consequential number in this literature is not an efficacy figure but a failure figure. First-line agents fail to adequately control tremor in an estimated 25% to 55% of people with essential tremor (Bruno et al., Cochrane, 2017). That is a wide interval, reflecting heterogeneous definitions of "adequate," but even its lower bound means a quarter of patients started on first-line therapy will need something more.

Two mechanisms sit behind that gap and are worth distinguishing. The first is genuine pharmacological non-response — the drug, adequately dosed, simply does not control the tremor. The second is tolerability-limited effectiveness: a drug that could work at a higher dose can't be escalated to it because of adverse effects, so its real-world ceiling falls below its efficacy ceiling. Both push effectiveness below the efficacy the trials report, which is why the practical failure rate a clinic sees can feel higher than "Level A" suggests.

The practical corollary is that the response gap is a timing question as much as a drug-selection one. Because roughly a quarter to a half of patients will not reach adequate control on the first agent, an escalation pathway should be anticipated at initiation rather than improvised after repeated failures: sequential first-line monotherapy, then combination, then the recognized second-line agents (topiramate, gabapentin, and benzodiazepines — Level B options whose evidence is more limited and mixed than the first-line pair's), and finally referral for procedural evaluation when medication is exhausted (Hopfner & Deuschl, Neurotherapeutics, 2020; Shanker, BMJ, 2019). Framing first-line therapy as the first rung of a defined ladder — rather than as the treatment — is what keeps a non-responding patient from spending months cycling doses before a specialist referral is considered.

Direct comparative evidence is surprisingly limited for two drugs used for decades. The most-cited head-to-head signal comes from Koller and Royse, in which primidone reduced tremor more than propranolol and did so at low doses (Koller & Royse, Neurology, 1986) — but this is a single study, and the guideline treats both as co-equal Level A agents rather than ranking them (Zesiewicz et al., Neurology, 2011). In contemporary practice the selection is therefore driven less by a decisive efficacy difference and more by the comorbidity and tolerability profile: propranolol's cardiorespiratory contraindications (asthma, bradyarrhythmia) push toward primidone, while primidone's sedation and notorious acute first-dose reaction push the other way. The two are also used in combination when neither alone suffices, and primidone retains efficacy in patients already on propranolol (Koller & Royse, Neurology, 1986).

Comorbidity frequently makes the selection for the clinician before efficacy does. A patient with coexisting hypertension, a migraine tendency, or a prominent anxiety component is a natural fit for propranolol, which addresses more than the tremor; a patient with asthma, COPD, a bradyarrhythmia, or brittle diabetes (where beta-blockade can mask hypoglycaemic warning signs) is steered toward primidone by exclusion. Age and cognition pull the other way — primidone's sedative burden is less forgiving in older or cognitively vulnerable patients, and its acute reaction can be alarming enough to end the trial on day one if the patient isn't warned. Because of that reaction, primidone is characteristically started very low (on the order of 12.5–25 mg at night) and titrated slowly, whereas propranolol is typically begun at a low daily dose and built toward the tremor-effective range over weeks. Neither drug's benefit is well predicted by serum concentration, so titration is guided by clinical tremor response and tolerability rather than levels (Koller & Royse, Neurology, 1986; Hopfner & Deuschl, Neurotherapeutics, 2020).

The distance between trial efficacy and clinic effectiveness in essential tremor is driven largely by tolerability. Primidone's acute first-dose reaction — nausea, vertigo, unsteadiness, sedation — is a well-recognized cause of early abandonment, which is precisely why the drug is initiated at very low, often bedtime, doses and titrated slowly. Propranolol carries dose-limiting fatigue, bradycardia, and its respiratory and cardiac contraindications. Each of these converts a proportion of patients who might have benefited into patients who discontinue, and none of it is captured by an efficacy point estimate. Broader reviews of managing essential tremor frame first-line therapy as a starting tier from which a substantial minority must be escalated rather than a reliable endpoint (Hopfner & Deuschl, Neurotherapeutics, 2020; Shanker, BMJ, 2019).

Long-term effectiveness is among the least well-characterized parts of this evidence base. Patients and clinicians frequently describe a first-line drug that "worked and then stopped," but the mechanism is usually disease progression outpacing a fixed dose rather than true tachyphylaxis, since essential tremor is progressive — dose adjustment or the addition of a second agent often restores control. A systematic review of pharmacological management and its long-term effect on quality of life underscores both that these drugs meaningfully affect quality of life and that sustaining benefit typically requires ongoing regimen adjustment rather than a single stable dose, while noting how limited the long-horizon data remain (Alharbi et al., Cureus, 2024). Tremor is, in this sense, a moving target, and "how well does first-line medication work" has a time axis the pivotal short-term trials never measured.

Several practically important questions remain under-answered by the current literature, and it is worth naming them plainly rather than implying more certainty than exists. There is no large, modern, head-to-head randomized comparison of propranolol and primidone to rank them or to define which patient phenotypes respond preferentially to each. Dose-optimization evidence is thin, and the Koller finding that low doses match high ones has not been extensively re-tested. Long-term effectiveness, discontinuation trajectories, and the real-world response rate under contemporary titration practice are poorly quantified. The reasonable reading is not that first-line therapy is weak — it genuinely helps a majority — but that the confidence intervals around exactly how well, for how long, and for whom are wider than the tidy "Level A, first-line" label suggests (Schneider & Deuschl, Neurotherapeutics, 2014).