Exercise-based interventions, potentially serving as either alternative treatments for depression or alongside medication and/or therapy, are recommended by the UK National Institute for Health and Care Excellence (NICE). Exercise is a broad term, and can range from slow guided movement such as yoga, to high-intensity strength or aerobic training, and numerous clinical trials for depression have been conducted over the past several decades. Understanding the effectiveness of different types of exercise in treating depression, and whether specific activities might be better suited to different individuals, is important for the development of both interventions and treatment guidelines.
Previous Mental Elf blogs have covered research establishing the effectiveness of exercise as a treatment for depression (Nedoma, 2023), including a meta-analysis finding exercise interventions to be “non-inferior to current first line treatments” (Heisel et al., 2023). As noted by Francesca Bentivega in her 2022 blog, although reductions in depression from exercise interventions are observed reliably, the mechanisms behind these changes are poorly understood, and may well differ between individuals and types of exercise (Bentivegna, 2022).
In their 2024 review, Noetel and colleagues aimed to identify the most effective types and amounts of exercise (referring to both the length of intervention and frequency of sessions) for treating depression, and possible factors that influence who is likely to respond to treatment (known as moderators).

Does the type, frequency, and intensity of exercise impact its mental health benefits?
Methods
Five databases were searched over a two-year period, identifying 18,658 papers, with 54 additional studies identified from previous reviews. After initial screening, 1,738 studies were assessed for eligibility, with 218 meeting the inclusion criteria and included in the analysis. These inclusion criteria were:
- Randomised controlled trials
- Longer than one week in length
- At least one exercise group
- Participants met clinical cut-offs for major depression
- Depression reported as an outcome
- Adequate data to calculate an effect size for each study arm.
The primary analysis comprised a multilevel network meta-analysis, which in principle allows for a comparison between different types of exercise across studies, with effect sizes presented as benefits beyond ‘active control’ conditions. Control conditions varied substantially between studies, for example: usual care, placebo tablet, stretching, educational control, or social support; for the purpose of analysis, the authors combined these conditions into a single ‘active control’ group.
Results
The ‘network geometry’ figure from the paper (reproduced below) illustrates the number of participants in each intervention arm (indicated by the size of the circles) and the number of direct comparisons between arms (indicated by the thickness of the lines). It shows notable differences in the frequency with which different interventions have been investigated, and that different exercise types have mostly been compared with ‘active control’; there are very few direct comparisons between different types of exercise, or with first-line treatments for depression (i.e. selective serotonin reuptake inhibitors (SSRIs) or cognitive behavioural therapy (CBT)).

The intervention groups included in the meta-analysis, with the size of the dots indicating the number of participants, and the thickness of the lines indicating the number of direct comparisons between study arms. [View full size graphic]
Hedges g scores (within-subject standardised mean differences, known as effect sizes) were calculated for each study arm separately (i.e., post- versus pre-intervention, where a more negative number indicates a greater reduction in depression), and aggregated across studies within each intervention category. Each intervention was then compared with ‘active control’ (summarised below).

Effect sizes from each intervention compared with ‘active control’. The yellow bars on the right indicate credibility ratings, based on both sample size and the authors’ judgment of study quality. [View full size graphic]
Conclusions
The main conclusions of the study are as follows:
- Multiple types of exercise proved effective as treatments for depression, for example walking or jogging, yoga, and strength training, with dancing showing particularly strong effects (albeit with lower certainty owing to the small sample size);
- More intense exercise was associated with a greater antidepressant effect;
- Benefits appeared to be similar for different baseline depression scores, and comorbidities;
- Age and gender might influence which exercise type is likely to be effective, although this needs to be confirmed in studies specifically designed to investigate individual differences.
The authors conclude that treatment guidelines around exercise for depression may currently be too conservative, and that tailoring exercise to individual patient characteristics might increase the chances of success.

Current exercise guidelines for depression may be too cautious. Personalising exercise plans to fit each individual could boost their chances of feeling better.
Strengths and limitations
The rigorous inclusion criteria and wide breadth of studies screened are important strengths of the review. Only randomised controlled trials including participants who reached clinical cut-offs of depression were analysed, meaning we can be confident that the results reported reflect the effects of interventions on participants with clinical depression.
The application of network meta-analysis is also a strength, as it potentially allows for the comparison of different types of intervention, despite the lack of many direct comparisons within studies. However, the type of “arm-based” network meta-analysis applied in this study is also a weakness, as unlike “contrast-based” analysis, it does not preserve the randomised aspect of comparisons within trials; therefore, caution must be applied when weighing up the effectiveness of different interventions in this manner, owing to the indirect nature of the comparisons. This limitation is particularly relevant to the comparisons between different types of intervention (e.g., exercise vs SSRIs), especially as the authors did not actually compare any of the different interventions statistically. Nonetheless, the results provide a useful illustration of which types of exercise are most commonly investigated, what they have been compared to, and preliminary indications of which ones work best. This lays the groundwork for future research to address understudied areas. Similarly, the rigour with which the authors assessed the risk of bias and study quality allows the identification of weak points within the existing literature.
One complication in interpreting the results is the use of a combined ‘active control’ condition, comprising usual care, placebo tablet, stretching, educational control, or social support. For example, presumably ‘stretching’ has several similar features to ‘yoga’ or ‘tai chi/qigong’, while ‘educational control’ could potentially be quite similar to ‘physical activity counselling’. These grey areas between the ‘active control’ and intervention groups, with the combined ‘active control’ group including interventions similar to those with both clinically important benefits (e.g., stretching and relaxation), and those below that threshold (e.g., physical activity counselling), could muddy the waters, potentially resulting in either under- or over-estimation of effect sizes.
Furthermore, as any comparison between interventions within the analysis is necessarily indirect (as few studies included more than one type of exercise), a high level of caution is necessary when drawing conclusions. This is particularly relevant for interventions with lower sample sizes and few direct comparisons, such as dance, as estimates of the difference between interventions will be imprecise. Similarly, the possible sex and age differences identified through exploratory moderation analyses are very preliminary and require further investigation, especially as analyses were based on the average gender mix or age in each study arm, rather than investigated at the individual participant level.
An additional and substantial limitation is the small sample size of most of the studies included in the meta-analysis (on average 100 participants), as well as clear evidence for publication bias (this refers to the fact that studies reporting significant results are more likely to be published). This raises the prospect that the results of the meta-analysis may not be accurate, as effect sizes for exercise interventions vs ‘active control’ will be over-estimated. However, it is worth noting that even when analysing only studies that reported a non-significant result, the authors still observed a significant effect when comparing exercise (across all types) with ‘active control’, albeit unsurprisingly with a much smaller effect size. To address this issue, we need large trials that evaluate the effectiveness of exercise interventions, ideally incorporating multiple intervention arms (which would allow for direct comparisons between different types of exercise).
A final limitation is similar to the point raised in the aforementioned blog “Aerobic exercise improves symptoms in students with major depression”, which emphasised our lack of understanding regarding the mechanisms through which exercise improves depression. Whilst this study lays out the evidence base supporting the implementation of exercise as a treatment for depression, the mechanisms behind its effectiveness remain unclear – especially given the wide range activities studied, which presumably could operate through different mechanisms. Additionally, the types of exercise showing promise varied considerably in terms of group/solo format, intensity, movement type, muscle groups used, and aerobic/anaerobic demands, raising important questions as to which of these factors are most important for treating depression, and underscoring the importance of mechanistic research. Although preliminary evidence of mediating factors, such as self-esteem, self-efficacy, and social connection exists (White et al., 2024), these factors have primarily been assessed using self-report questionnaires, and the cognitive, neural and physiological mechanisms driving these changes are far less well understood. To address this question, mechanistic randomised controlled trials are required, which would involve taking cognitive, neural and physiological measures before, during and after exercise and control interventions, in the context of a randomised design.

A wide variety of activities can help treat depression, ranging from yoga and tai chi to high-intensity strength and aerobic training.
Implications for practice
Although physical activity is recommended by the World Health Organisation (WHO) and included in the UK NICE guidelines for depression, arguably it is not ‘prescribed’ at a frequency commensurate with its supporting evidence base. In addition to summarising the existing literature supporting the use of exercise interventions for depression, this paper provides preliminary evidence for the potential utility of tailored exercise recommendations based on individual characteristics, as well as suggesting that more intense exercise is likely to be more effective.
Keeping exercise in mind as a viable treatment option for depression is advisable, and emphasising its potential not just as an alternative, but also potentially as an addition to first-line treatments, may be important to encourage more uptake. Community-based activity classes specifically designed for people with depression are extremely uncommon in the UK, therefore increasing funding for exercise-based interventions within mental health or primary care services could lead to greater awareness and reduce existing barriers to both prescription and attendance.
Importantly, few studies in the meta-analysis obtained follow-up data for more than three months, and therefore the longer-term impact of exercise on depression remains uncertain – even when exercise is effective, it is likely that if individuals stop exercising then the risk of relapse will be high. Therefore, a key goal in terms of implementation in services should be to encourage people to maintain an exercise routine in the longer term, after the end of a programme of classes. For example, this might be achieved by designing interventions that can be implemented more practically into people’s lives, and routines such as the concept of exercise ‘snacks’ (Thøgersen-Ntoumani et al 2024).
Another way of facilitating a cultural shift towards greater use of exercise in mental health settings is through understanding the mechanisms driving positive change. For example, the LIFE Trial currently taking place at University College London (UCL), is investigating the biological, neural, and psychological processes changed by exercise in depression. If such research does yield improved mechanistic understanding, it could not only increase patient and healthcare professionals’ willingness to consider it, but also provide further indications as to how exercise interventions can be tailored to individuals.

Despite being included in NICE and WHO guidelines, exercise is not ‘prescribed’ enough for people with depression.
Statement of interests
The authors are part of the study team of the LIFE Trial mentioned in the ‘Implications for Practice’ section, but otherwise declare no conflicts of interest.
Acknowledgements
Sincere thanks to the rest of the UCL LIFE study team (Prof Glyn Lewis, Prof Mark Hamer, Dr. Emily Hird, Dr. Elle Newton, Ashley Slanina-Davies, Jehanita Jesuthasan) for their comments and feedback that helped shape this blog, to Prof Deborah Caldwell for kindly answering several statistical queries, and to Dr Michael Noetel, the first author of the discussed paper, for his timely and detailed responses to our enquiries during the writing process.
Links
Primary paper
Noetel M, Sanders T, Gallardo-Gómez D, Taylor P, Del Pozo Cruz B, van den Hoek D, Smith JJ, Mahoney J, Spathis J, Moresi M, Pagano R, Pagano L, Vasconcellos R, Arnott H, Varley B, Parker P, Biddle S, Lonsdale C. (2024) Effect of exercise for depression: systematic review and network meta-analysis of randomised controlled trials – PubMed. British Medical Journal 2024, 384 q1024.
Other references
Francesca Bentivegna (2022) Aerobic exercise for major depression: the role of reward processing and cognitive control. https://www.nationalelfservice.net/treatment/exercise/aerobic-exercise-major-depression/
Ross Nedoma (2023) Exercise for depression: an evidence-based treatment option. https://www.nationalelfservice.net/treatment/exercise/exercise-for-depression-an-evidence-based-treatment-option/
Heissel, A., Heinen, D., Brokmeier, L. L., Skarabis, N., Kangas, M., Vancampfort, D., Stubbs, B., Firth, J., Ward, P. B., Rosenbaum, S., Hallgren, M., Schuch, F. (2023). Exercise as medicine for depressive symptoms? A systematic review and meta-analysis with meta-regression. British Journal of Sports Medicine 2023 57 1049-1057.
Thøgersen-Ntoumani, C., Grunseit, A., Holtermann, A., Steiner, S., Tudor-Locke, C., Koster, A., Johnson, N., Maher, C., Ahmadi, M., Chau, J. Y., & Stamatakis, E. (2024). Promoting vigorous intermittent lifestyle physical activity (vilpa) in middle-aged adults: an evaluation of the movsnax mobile app. BMC public health, 2024 24(1) 2182.
White, R. L., Vella, S., Biddle, S., Sutcliffe, J., Guagliano, J. M., Uddin, R., Burgin, A., Apostolopoulos, M., Nguyen, T., Young, C., Taylor, N., Lilley, S., Teychenne, M. (2024). Physical activity and mental health: a systematic review and best evidence synthesis of mediation and moderation studies. International Journal of Behavioural Nutrition and Physical Activity, 2024 21(134)