Drawing on evidence from case reports, a medication treatment trial, clinicopathological, neurophysiological, and imaging studies, this systematic review of 25 articles demonstrates that tremor is present in up to 97.5% of patients with MSA, with notable differences depending on phenotype and tremor type.

Our results align broadly with earlier reports (Tison et al. 2002; Kaindlstorfer et al. 2013; Mailankody et al. 2017). However, we identified a much wider prevalence range, with overall tremor prevalence ranging from 10% to 97.5%. Some studies even reported a prevalence of action/postural tremor in MSA that exceeded those of earlier cohorts of patients with PD, in whom tremor occurred in approximately 75% of patients (Koller et al. 1989; Gironell et al. 2018).

Our updated synthesis indicates a prevalence of rest tremor in approximately one in three to four patients with MSA-P (~ 26–38%), which is higher than observed in patients with MSA-C (~ 11–23%), and lower than previously reported (Kaindlstorfer et al. 2013). Classic pill-rolling tremor was only reported in two articles (Miki et al. 2019; Pradhan and Tandon 2023), with higher prevalence than reported earlier by Wenning et al. (Wenning et al. 1994). However, although isolated cases of pill-rolling rest tremor have been reported in patients with autopsy-confirmed MSA (Miki et al. 2019), the reports by Pradhan and Tandon rely solely on clinical diagnosis. In both studies, the identification of pill-rolling tremor was based on clinical observation by a single neurologist, without interobserver validation or neurophysiological confirmation. Overall, rest tremor in MSA occurred significantly less frequently than in PD, where half of patients experience rest tremor as initial disease manifestation (Kipfer 2011). Rest tremor does not reliably distinguish MSA from PSP (6–44%) or corticobasal syndrome (CBS: 19–21%), due to overlapping prevalence data (Mailankody et al. 2017).

Consistent with prior findings (Kaindlstorfer et al. 2013; Mailankody et al. 2017), we observed that postural tremor is the most common tremor type in MSA. However, while previous findings reported a predominance of postural tremor in MSA-P compared to MSA-C (Yabe et al. 2006; Wenning et al. 2013; Kaindlstorfer et al. 2013; Mailankody et al. 2017), when taking lower limb postural tremor into account, our results contradictorily suggest that on average, postural tremor with or without jerky component, has a higher occurrence in MSA-C.

Intention tremor is previously reported in less than one-third of patients with MSA-P and roughly one-third to 45% of patients with MSA-C (Kaindlstorfer et al. 2013). However, other studies rarely reported intention tremor, and when mentioned, the specific MSA subtype was not identified (Miki et al. 2019).

Importantly, re-emergent tremor, commonly observed in 4–66% of patients with PD (Mailankody et al. 2017), appears to be absent in MSA, according to a previous study and our current synthesis of evidence (Kaindlstorfer et al. 2013).

Notably, orthostatic tremor and orthostatic myoclonus, which have not been reported at all in earlier reviews, gained recent evidence in our review, as low-frequency (3-Hz) pseudo-orthostatic tremor occurred in up to 38.3% of patients with MSA-P and up to 95.7% of patients with MSA-C (Wang et al. 2025). Similarly, orthostatic myoclonus at 3-Hz was identified in three MSA-P cases (Gasca-Salas et al. 2013). While tremor of the legs, head, chin, lips, and tongue has previously been described as uncommon in MSA (Kaindlstorfer et al. 2013), our synthesis showed that vocal tremor was reported with a prevalence of 54% (Rusz et al. 2015), more common in MSA-P but also in patients with MSA-C (Rusz et al. 2015; Hlavnička et al. 2020; Mir et al. 2024).

Neuroimaging evidence (Gallea et al. 2016) substantiates these findings by demonstrating that patients with orthostatic tremor showed greater atrophy in cerebellar lobule VI, particularly lower grey matter volumes, correlating with more severe postural instability. This supports the hypothesis that cerebellar degeneration contributes to the development of orthostatic tremor, and may explain its higher prevalence in MSA-C compared to MSA-P. In line with this, Wang et al. (Wang et al. 2025) found that the 3-Hz pseudo-orthostatic tremor is not only frequent in MSA-C but absent in a large cohort of patients with PD. This suggests a potential diagnostic utility of low-frequency pseudo-orthostatic tremor as a clinical marker for distinguishing MSA from PD.

Neurophysiological data indicate that a slow 3-Hz pseudo-orthostatic tremor is characteristic of MSA-C, whereas MSA-P more frequently exhibits a higher-frequency tremor or irregular myoclonic bursts, with only mild or absent 3-Hz tremor. The 3-Hz pseudo-orthostatic tremor observed in MSA-C lies well below the frequency range of classic primary orthostatic tremor (~ 13–18-Hz) and even below that of typical pseudo-orthostatic tremor (~ 6–7-Hz) (Thomas et al. 2007). When tremor occurred in MSA-P, it tended to be faster than in typical PD: resting tremor averaged approximately 5.5-Hz, and postural or weight-holding tremor reached up to 7-Hz, exceeding PD frequencies across all activation states (Su et al. 2020). While the limbs, especially the upper limbs, were the most frequently affected body part in both MSA and PSP tremor, occasionally extending to the lower limbs and face (Fujioka et al. 2016), pseudo-orthostatic tremor has not been convincingly documented in other atypical parkinsonian syndromes. For example, in PSP only isolated cases of primary orthostatic tremor have been reported (Maréchal et al. 2024). In Corticobasal degeneration, where postural tremor is the most prevalent tremor type, the frequency is generally higher (6–8-Hz) and tremor appears more jerky and irregular than that seen in PD (Mahapatra et al. 2004), however, orthostatic tremor has not been reported. These observations indicate that a low-frequency pseudo-orthostatic tremor of the lower limbs may serve as an important differential diagnostic feature. Accordingly, objective characterization with neurophysiological investigations are crucial. This allows for instance the detection of pure postural tremor, indicating atypical parkinsonism, which appears immediately on assuming posture with a lower amplitude, higher frequency, and a broader spectral peak, compared to PD re-emergent tremor, and often displays a jerky, irregular pattern with short (< 50 ms), synchronous bursts (Schwingenschuh et al. 2025).

Given the importance of treatment response in differentiating PD from atypical parkinsonian syndromes, we synthesized the available literature on tremor response to levodopa and other treatments. According to current MSA diagnostic criteria (Wenning et al. 2022), clinically established MSA is characterized by poor levodopa responsiveness (< 30% improvement on the MDS-sponsored revision of the Unified Parkinson’s Disease Rating Scale Part III with up to 1000 mg levodopa). We found heterogeneous and generally modest tremor responsiveness to levodopa in MSA. Natural history studies reported levodopa response in 42.5%–56.7% of patients with MSA-P and 12.9%–25% with MSA-C, with average response duration of 3.3–3.5 years and 2.6–3.3 years, respectively (Wenning et al. 2013; Low et al. 2015). In line with previous studies (Wenning et al. 2013), we observed that patients with MSA-P are more likely to benefit from levodopa than MSA-C. Non-levodopa interventions, such as botulinum toxin, benzodiazepines, piracetam, targeted combinations and focused ultrasound thalamotomy have been recently explored, suggesting potential amelioration of specific tremor phenotypes in individual cases, but high-quality efficacy evidence is still lacking (Brum et al. 2016; Balint et al. 2021; Sinai et al. 2024; Bhattacharjee et al. 2024). In single cases, clonazepam lead to mild improvement in myoclonus (Shindo et al. 2014; Hwang et al. 2019), again without high-quality data. We could not confirm previous findings on levodopa-responsiveness in low-frequency orthostatic tremor (Thomas et al. 2007).

Although no deep brain stimulation (DBS) study met inclusion criteria, external case series and retrospective analyses describe only transient or limited motor benefits with neurobehavioral complications (Artusi et al. 2022; Badihian et al. 2022), arguing against routine DBS in atypical Parkinson’s syndromes.

Compared to patients with PD, patients with MSA showed significantly less motor response to levodopa, while PSP and MSA did not differ in levodopa responsiveness (Ye et al. 2024). As previously pointed out (Kaindlstorfer et al. 2013), the results of randomized controlled trials on treatment responses in MSA tremor are inconsistent and only limitedly proven.

Beyond treatment responsiveness, MSA phenotypes also differ in terms of their geographical occurrence, with MSA-P diagnoses occurring more often in European and US studies and MSA-C in Asian studies, which aligns with earlier findings (Yabe et al. 2006; Wenning et al. 2013). More specifically, we found that Asian studies observed postural tremor with higher frequency in MSA-C than in MSA-P (Li et al. 2018, 2021), whereas US (Low et al. 2015) and European studies (Miki et al. 2019), found the opposite pattern. These findings are consistent with previous ones (Mailankody et al. 2017).

Since the updated diagnostic criteria for MSA (Wenning et al. 2022) require at least one indicative marker in addition to the central clinical features, the diagnostic relevance of specific tremor phenotypes in differentiating MSA from other parkinsonian syndromes has increased. Therefore, the systematic and precise documentation of tremor in patients with MSA is essential. The International Parkinson and Movement Disorder Society’s two-axis tremor classification (Bhatia et al. 2018) offers a practical framework for this purpose, which also helps create well-defined patient subgroups that can be reassigned if the clinical picture changes (Bhatia et al. 2018). Accordingly, tremor phenotypes that have been proposed as supportive of MSA should be systematically sought and characterized electrophysiologically in future cohorts, as their presence can significantly improve diagnostic specificity in both clinical practice and research. For instance, tremor and myoclonus can be distinguished based on clinical neurophysiological investigation of agonist-antagonist contraction pattern and rhythm regularity (Schwingenschuh et al. 2025). However, an important knowledge gap concerns tremor severity and its clinical relevance in MSA, including how often tremor is the presenting symptom or tremor treatment is requested by patients. Across included studies, tremor severity was seldom quantified and validated tremor-specific scales were almost never used, with the exception of a pilot interventional case series (Sinai et al. 2024). Most reports relied on global motor scores without tremor subscores (Su et al. 2020; Ye et al. 2024), precluding inferences about tremor burden. Likewise, tremor as the initial complaint was only noted in isolated MSA-P cases, and no study systematically assessed whether tremor severity or functional impact drove treatment seeking. This highlights the need for standardized tremor-specific severity and impact assessments in future cohorts.

Such precise characterization is particularly relevant given the growing emphasis on early and accurate diagnosis to inform patient-centered care. Recognizing this, the ongoing MeDeMSA Care study (Fanciulli et al. 2025) aims not only to alleviate symptomatic burden and attenuate quality-of-life decline but also to tailor management strategies according to individual healthcare preferences in patients with MSA.

This review, which is the first that systematically analysed tremor reports in MSA, is characterized by its comprehensive methodology, including diverse study types, ranging from experimental to observational designs and a systematic quality assessment, therefore providing a broad and integrative overview of the current evidence base.

However, as all systematic reviews, our study has several limitations. First, because multiple system atrophy is a rare disease, only a limited number of larger cohort studies were available, constraining the overall evidence base. In rare diseases, smaller studies or studies reporting null or atypical findings may be less likely to be published. This selective publication may bias the available evidence base and lead to distorted prevalence estimates.

Second, the included studies were heterogeneous in study design patient populations, diagnostic criteria and outcome measures, and often lacked standardized, comparable assessments. This limited cross-study comparability and led us to present a narrative summary rather than conduct a quantitative meta-analysis.

Third, only 112 (18.9%) of included MSA patients had an autopsy-confirmed diagnosis, such that the reliance on clinical diagnostic criteria may have led to misclassification, while the uneven distribution of autopsy-confirmed cases across studies, and the lack of stratification by MSA phenotype and tremor types prevented meaningful comparative subanalysis of these cases.

Fourth, when synthesizing prevalence data, we applied a minimum sample size threshold of n < 40 patients with MSA to reduce the risk of imprecise prevalence estimates. This criterion was not applied to other domains addressed in this review, where smaller and exploratory studies were considered informative and therefore included. Consequently, some smaller prevalence studies were excluded, which may have affected the completeness of prevalence estimates and should be considered when interpreting these results.

Finally, methodological limitations must be acknowledged. The literature search was conducted primarily in PubMed, supplemented by reference tracking and targeted web searches. While this strategy likely captured most clinically relevant studies, eligible publications indexed exclusively in other databases may have been missed. Moreover, restricting the search to German- and English-language publications, introduces a potential language bias. Limiting inclusion to studies published from 2013 onward may also have excluded earlier relevant work, although this decision was made to ensure alignment with contemporary diagnostic criteria and assessment methods.

In summary, tremor is a common but heterogeneous feature of MSA. Action tremor predominates and occurs in a high proportion of patients with MSA-C in the lower limbs and in a high proportion of MSA-P in the upper limbs, while rest tremor is less frequent. For differential diagnosis, low-frequency pseudo-orthostatic tremor and orthostatic myoclonus are particularly relevant, as they are common in patients with MSA-C but rare in patients with PD and other atypical parkinsonian syndromes. Levodopa response is generally modest in MSA compared to PD, and evidence for other treatment remains limited and of low quality. Overall, there is a notable lack of systematic phenotyping of tremor and electrophysiological characterization in MSA. Broader adoption of the MDS two-axis tremor framework and routine, standardized documentation of tremor phenomenology may facilitate earlier early detection, improve diagnostic precision, and support more targeted research.