Overtraining: effects & state of evidence — what’s supported and what isn’t

Overtraining is more than “simply doing too much training.” Research often separates non-functional overreaching (short- to mid-term, potentially reversible) from the broader and more problematic overtraining syndrome (OT syndrome). Because studies use different sports, markers, and timeframes, making a generally applicable statement about effects is difficult to support.

In this article, I sort out which effects and markers (e.g., HRV) are treated as plausible in the reviews—and where the data remain limited. The goal is better load management, not using supplements as “compensation.”

Recognizing overtraining: why “too much” isn’t automatically proven

Overtraining is not automatically the same as “more training.” The evidence base uses inconsistent definitions and operationalizations, so performance declines, recovery problems, and lab/functional tests are not cleanly comparable. Especially for single markers like HRV, interpretation depends heavily on the sport, measurement design, and the load window.

In practice, “too much” feels intuitive: if performance drops and recovery worsens, it suggests the workload is too high. Scientifically, however, this logic is rarely tested directly as “overtraining caused by dosage X.” Instead, many studies use phase-based terms such as non-functional overreaching and overtraining syndrome, but differ in when measurements are taken and which criteria (performance, questionnaires, lab values, subjective symptoms) define the transition.

That is one reason the strength of evidence for hard, universally valid “cause → effect” statements is limited: the same symptoms can arise from sleep loss, infectious load, psychological stress, or inappropriate training periodization. Reviews discuss exactly this heterogeneity problem and propose assessing overloading via repeated patterns across multiple training cycles rather than a single measurement.

One example for the marker question is HRV. The systematic review in soccer players shows recurring patterns linked with overload phases, but: findings are sport-specific and not transferable 1:1 to every sport. The evidence is therefore framed more as a “potential signal” rather than a universal diagnostic test. (Lipka et al., 2025, PMID 40405528).

If overload persists across multiple training cycles, what begins as “too much load” can evolve into a problematic pattern that reviews discuss as part of a recovery disorder and/or a persistent maladaptive response. (Valdesalici et al., 2026, PMID 41580212)

Evidence hierarchy: what systematic reviews, reviews, and primary data actually provide

Systematic reviews are best for a reliable overall view. For HRV and overtraining, the evidence base is especially focused within a systematic overview. For psychological and cognitive aspects, systematic evidence also exists, but included studies are more heterogeneous—so effect estimates can vary between studies. Frameworks help with monitoring design, but they rarely provide direct RCT effect sizes.

The key question is: What kind of evidence answers your question? If you want to know whether a marker like HRV can serve as a signal under overload, systematic reviews are often the next sensible step. There is a systematic overview that compiles HRV within the context of overtraining/overload in soccer players. It positions HRV as a possible signal while emphasizing limits: sport, measurement protocols, and study designs are not identical. (Lipka et al., 2025, PMID 40405528)

For psychological and cognitive outcomes, evidence is also summarized via a systematic review. This work maps the relationship between non-functional overreaching / overtraining syndrome and psychological/cognitive function—again with heterogeneity in participant groups, duration, measurement instruments, and definitions, which makes it difficult to derive fixed cutoffs or “doses.” (Valdesalici et al., 2026, PMID 41580212)

Frameworks and systems-level thinking are a separate category. Articles that describe biomarkers as building blocks for a monitoring system can be useful for deriving practical decision logics (e.g., layered monitoring, temporal signals instead of single values). However, they often do not provide the type of evidence you would need to conclude “marker X improves after intervention Y” with an effect size from a large RCT. (Mânescu et al., 2026, PMID 42074239) (Mânescu et al., 2026, PMID 42074313)

For real life this means: if you use a measurement variable, treat it as part of a bundle of decisions, not as the only trigger. Otherwise you run into the same issue that overtraining discussions sometimes “hide”: you see a signal, but not necessarily the correct cause.

Further, if you want to use markers as decision logic, a systems view emphasizes multi-layered and time-linked assessment rather than isolated “instant proof.” (Mânescu et al., 2026, PMID 42074313)

Lifestyle levers as the core: sleep, recovery, and load management first

If you want to avoid overtraining, sleep, recovery, and load management are the most evidence-adjacent levers. Supplements can at most play a supportive role, but the fundamentals lie in the overall stress balance—not in a “compensation pill.”

Why put sleep so far forward? Not because “sleep cures everything,” but because sleep is tightly linked to recovery, readiness to perform, and physiological regenerative processes. A narrative review describes mechanisms through which sleep influences sports performance and recovery, and frames sleep as a priority before additional measures. (Barreira et al., 2025, PMID 41217703)

Load management is also increasingly framed in newer models as total stress. The 4Rs concept (as a framework) aims to better evaluate allostatic load level in athletes instead of treating overload purely as training volume. This shifts perspective: overload may result from the sum of multiple resource/stress sources (training stress, recovery stress, energy availability, etc.) rather than only the hard session “on Tuesday.” (Bonilla et al., 2025, PMID 40566521)

If you want to use biomarkers (e.g., HRV), systems-level thinking emphasizes that you need practical recovery windows in your training plan, rather than trying to “make up” monitoring decisions later through supplementation. This is an important methodological point: monitoring is useful when you have a real, actionable adjustment you can make soon. (Mânescu et al., 2026, PMID 42074313)

Practically, this means:

• Prioritize sleep quality and consistency first. (Barreira et al., 2025, PMID 41217703)
• Actively shape recovery phases and training monotony/variability.
• Use markers like HRV as a warning light for de-loading/timing adjustments, not as justification to “keep pushing and refill later.” (Lipka et al., 2025, PMID 40405528)

If you also want to understand why effects in studies are often small to moderate—and why monitoring can still be worthwhile—look at the topic effect size: Effektstärke verstehen: Wirkung & Studienlage von 1–2 Hebeln.

What studies suggest about “effects”: HRV, biomarkers, and mental consequences

The strongest evidence within the listed studies supports HRV as a potential signal for overtraining/overload in certain athletic contexts. For biomarkers, convincing systems-level frameworks exist, but effect sizes are often less robust for each marker. For psychological and cognitive consequences, a systematic review suggests a relationship; however, the strength varies across studies and does are not provide universal cutoffs.

HRV: signal, but not a “universal diagnosis code”

The systematic review on HRV and overtraining in soccer players concludes that HRV in this context is discussed as a signal for overload/overtraining-like states. But: studies are sport-specific and methodologically heterogeneous. Therefore, moving from “HRV changes” to “HRV reliably tells you overtraining in every sport” is not well supported. (Lipka et al., 2025, PMID 40405528)

Biomarkers: systems view, not a single-marker miracle

At the molecular level, biomarkers are described in a systems framework as part of training response and as monitoring components. These works help you understand how biomarkers should be interpreted over time together with contextual variables. At the same time, framework articles typically provide fewer direct, intervention-related effect sizes (“take marker X and overtraining improves by Y%”), and more often structure thinking and the study-design logic. (Mânescu et al., 2026, PMID 42074239)

In a second framework, the idea is extended: biomarkers as temporal signals integrated across multiple levels into a decision (recovery, overload, adaptation). This directly addresses your core issue: single measurements are often too transient or too nonspecific to establish a clear cause on their own. (Mânescu et al., 2026, PMID 42074313)

Mental and cognitive consequences: direction yes, strength variable

For the psychological/cognitive side, a systematic review summarizes how non-functional overreaching and overtraining syndrome influence psychological and cognitive function in elite athletes. It emphasizes that study conditions are heterogeneous, so direction and/or effect magnitude can vary between studies. This matters because it implies: “It can feel like overtraining” can be true, but “the effect size is always the same” cannot be derived reliably from this evidence. (Valdesalici et al., 2026, PMID 41580212)

If you’re looking for support from training biology, it can also help to place hormonal axes in context—but as a review perspective, not a direct overtraining intervention logic. (Cano et al., 2016, PMID 27348623)

Overload vs. overtraining: practical framing that respects the evidence base

In the literature, non-functional overreaching is often interpreted as a reversible intermediate phase between adaptive stress and a more problematic course. Following this logic, markers and symptoms can change depending on the phase. That is why the evidence supports time-based, multi-layered monitoring rather than deciding based on a single signal.

Why is the distinction so central? Because the term “overtraining” works like a switch in everyday language (“present or not present”). In studies, reality is more often dynamic: load → adaptive stress response → possible compensation → if the load remains unfavorable: recovery disorder/disease-like states or persistent performance problems.

The systematic review on psychological and cognitive consequences makes exactly this point: non-functional overreaching is often positioned in the literature as a phase between adaptive stress and a problematic course. This makes it hard to use a single marker as a “diagnosis,” because it may respond differently across time windows. (Valdesalici et al., 2026, PMID 41580212)

Frameworks therefore suggest treating overload as a temporal pattern and making decisions across multiple levels: performance, subjective/psychological signals, sleep/recovery parameters, plus possible biomarkers as additional evidence. (Mânescu et al., 2026, PMID 42074313)

The 4Rs framework concept also supports a multidimensional view of allostatic load. That is practically helpful: you can recognize that an athlete is “training,” but simultaneously has too little energy/recovery/stress management—making overload more likely. (Bonilla et al., 2025, PMID 40566521)

A practical way to handle the evidence base would therefore be:

• Treat “overtraining” first as a warning hypothesis, not a label.
• Use a decision protocol: if patterns (not single values) shift toward a recovery disorder, adjust training/sleep/load and observe the response over days to weeks.
• Avoid overinterpreting isolated blood values or a single HRV measurement point, because the systems view warns against precisely that. (Mânescu et al., 2026, PMID 42074313)

If you also want to know how cognitive and psychological parameters can be measured in practice: the systematic review provides an overview, but no “universal cutoffs.” (Valdesalici et al., 2026, PMID 41580212)

Study overview: how well different markers and endpoints are supported

HRV is the most clearly developed as a potential monitoring signal within the available evidence, but within specific contexts. Psychological/cognitive endpoints are also summarized systematically, yet effect strengths are variable and cutoffs are not reliably derivable. Biomarker frameworks mainly support monitoring design and temporal decision logic rather than clear, marker-based efficacy evidence.

If you want to interpret your own measurements, the core question is: Has that marker already been used in the studies as “signal-bearing” in a similar context? For HRV, the answer within the list is “best,” but still not universal. (Lipka et al., 2025, PMID 40405528)

For mental/cognitive endpoints, evidence is synthesized, but transferability depends strongly on the target population and the measurement instrument. Therefore, in practice it’s smarter to look for recurring patterns and parallel signals. (Valdesalici et al., 2026, PMID 41580212)

And for biomarkers: frameworks help you build monitoring so you can derive decisions (e.g., temporal windows, multi-layer criteria). However, they do not automatically provide the type of “efficacy evidence” you would need for a specific intervention per marker. (Mânescu et al., 2026, PMID 42074239) (Mânescu et al., 2026, PMID 42074313)

What to take away (Bottom Line)

• Overtraining is rarely tested in studies as a single cause → effect; terms, phases, and measurement designs are heterogeneous, so universally applicable effect claims are limited. (Valdesalici et al., 2026, PMID 41580212)
• HRV is a potential monitoring signal (within specific contexts), but not a universal diagnosis; transferability is limited. (Lipka et al., 2025, PMID 40405528)
• The most useful lever is lifestyle: sleep, recovery, and load management (allostatic total stress) before supplements. (Barreira et al., 2025, PMID 41217703) (Bonilla et al., 2025, PMID 40566521)
• Use biomarkers only as part of a multi-layer, time-oriented decision system, not as a “single-value verdict.” (Mânescu et al., 2026, PMID 42074313)
• If you want to infer “effects,” check the evidence level first: systematic reviews give direction, frameworks give design—but they do not automatically translate into intervention-related effect sizes.