What blood tests are worth running if you train hard, and how to read them

If you train seriously, the panel worth paying for is: a complete blood count with reticulocytes (CBC with retic), a comprehensive metabolic panel, ferritin plus transferrin saturation plus soluble transferrin receptor, high-sensitivity CRP, creatine kinase, 25-hydroxyvitamin D, a full thyroid set (TSH, free T4), ferritin-adjacent B12 and folate, a lipid panel with ApoB, and (if male) total and free testosterone with SHBG, or (if female) a cycle-timed estradiol, progesterone, FSH and LH. Run it three to four times a year at the same time of day, in the same fasted and hydrated state, at the same lab, and interpret each result against your own prior values rather than the population reference interval. A single draw tells you almost nothing about your training. Four draws over a year, plotted, tell you a great deal.

That last point is the whole argument. The published reviews of biomarker monitoring in athletes converge on the same conclusion: individual baselines and serial measurement, not one-off comparison to laboratory reference ranges, are what make blood work informative in trained populations ¹. Season-long studies in team sport show that markers move coherently with training load over weeks, which is the signal you want, and that the between-athlete scatter at any single time point is wide enough to swamp it ².

What belongs on the panel and why

The panel above is organized around four questions you can answer from blood, plus one category of general health screening you should run anyway.

Oxygen carrying capacity and iron status come first, because iron deficiency is the most common correctable finding in endurance athletes and because it is genuinely invisible on a standard CBC until late. Order hemoglobin, hematocrit, MCV, reticulocyte percentage and absolute count, reticulocyte hemoglobin equivalent (Ret-He, the hemoglobin content of newly made red cells, an early marker of iron-restricted erythropoiesis), ferritin, serum iron, transferrin, transferrin saturation, and soluble transferrin receptor (sTfR). Ferritin alone is a trap because it is an acute-phase reactant: it rises with inflammation, so a ferritin of 45 µg/L drawn 36 hours after a hard session with a CRP of 8 mg/L is not the same number as a ferritin of 45 µg/L at rest. Always pair ferritin with hs-CRP and read them together. sTfR is not an acute-phase protein, which is why it helps disambiguate.

Tissue damage and recovery status come second. Creatine kinase (CK) and lactate dehydrogenase rise after eccentric loading, with CK typically peaking 24 to 48 hours post-exercise and sometimes reaching several thousand U/L in an untrained or unusually loaded state. Studies of recovery in soccer players show CK and inflammatory markers tracking the load and the recovery interval, which makes them useful for answering “did I recover from that block?” and useless for answering “what is my baseline?” unless the draw is well separated from training ³. Urea and creatinine add a crude read on protein turnover and renal handling, but both are sensitive to hydration and protein intake.

Endocrine and metabolic context come third: total and free testosterone with SHBG, cortisol (morning, since the diurnal swing is larger than any training effect you are looking for), IGF-1, TSH and free T4, fasting glucose and insulin, HbA1c, and a lipid panel with ApoB. The testosterone-to-cortisol ratio is often marketed as an overtraining index and behaves poorly as one in individuals, largely because both analytes have large within-day and within-week variation. Treat it as one input among several, not as a verdict. Micronutrient status is fourth: 25(OH)D, B12, folate, magnesium, and (if you have reason) zinc.

The broader framework here matters more than the specific list. Recent work reviewing how to interpret routine blood chemistry alongside metabolomic profiling in athletes argues explicitly for reading markers in context, as a panel responding to a known load history, rather than as independent flags ⁴. That is the posture to adopt.

The reference interval is not your reference

A population reference interval is constructed to contain the central 95% of a reference population. Your own analyte has a within-subject biological variation (CV_I) that is usually much smaller than the between-subject variation (CV_G), which is exactly why “in range” can hide a real change and “out of range” can be nothing.

The arithmetic you want is the reference change value:

RCV = 2^0.5 * Z * sqrt(CVa^2 + CVi^2)

where CVa is analytical imprecision from your lab’s method validation and Z is 1.96 for 95% confidence in a bidirectional change. For ferritin with CVa near 5% and CV_I near 15%, RCV comes out around 44%: a move from 60 to 50 µg/L is noise, a move from 60 to 30 µg/L is real. For hemoglobin, where CV_I is roughly 3%, RCV is closer to 9 or 10%, so small hemoglobin changes are meaningful in a way small ferritin changes are not. Compute RCV per analyte once, store it next to the analyte in your schema, and let it gate your alerts. This single step removes most of the false signal that makes athlete blood testing feel noisy.

Timing rules that make the numbers comparable

Preanalytical variation is the largest error term in athlete blood work, and it is fully under your control.

Draw in the morning after an overnight fast of 10 to 12 hours, with water permitted. Do not extend the fast to make the numbers look better: controlled water-only fasting shifts glucose, lipids, and other metabolic markers substantially within 24 hours, so a 24-hour fast and a 12-hour fast are not interchangeable baselines ⁵. Keep the pre-draw window free of hard training for at least 48 hours, and ideally 72 if you are trying to establish a baseline rather than measure recovery. Sit upright for 15 minutes before venipuncture, because posture alone moves hemoglobin and hematocrit by several percent through plasma volume shift. Hydrate normally the evening before and the morning of.

Because plasma volume is the dominant confounder for every concentration you measure, compute it. The Dill and Costill correction, using hemoglobin and hematocrit from two draws, gives percent change in plasma volume:

dPV% = 100 * ((Hb_1/Hb_2) * ((100 - Hct_2)/(100 - Hct_1))) - 100

If dPV is −8% between draws, a 7% rise in total protein or albumin is a hydration artifact, not a finding.

Avoid finger-prick dried blood spot kits for anything quantitative where you care about small changes. The hematocrit of the spot changes how blood spreads on the card and how much analyte a fixed punch contains, producing a bias that scales with hematocrit and remains a real limitation of the format ⁶. Athletes, whose hematocrit ranges wider than the general population’s, are precisely the group this hurts.

Hematocrit, hemoglobin mass, and the passport arithmetic

Endurance athletes do not reliably have high hematocrit. Plasma volume expansion from training often lowers measured hematocrit even as total hemoglobin mass rises, which is the classic “sports anemia” artifact. Hematocrit is a concentration, and what carries oxygen is total hemoglobin mass.

Higher hematocrit is also not monotonically better. Whole blood viscosity rises steeply with hematocrit, and the hematocrit that maximizes oxygen transport depends on vessel geometry and flow conditions rather than sitting at some universal maximum ⁷. Very high values raise viscosity faster than they raise carrying capacity.

If you want a quantity that is more stable than hematocrit, borrow the anti-doping arithmetic. The OFF-score, used in the Athlete Biological Passport, combines hemoglobin and reticulocyte percentage:

OFF = Hb_g_per_L - 60 * sqrt(retic_percent)

Serial OFF-scores compress plasma-volume noise and respond to changes in erythropoietic drive. Plotting your own OFF-score across a season is more informative than plotting hematocrit. A value that sits far outside your own established range is a reason to see a physician, not a reason to self-interpret.

Getting the data out in a computable form

Insist on structured results. Most large reference laboratories can deliver HL7 v2 ORU^R01 messages or a FHIR Observation bundle through a patient or provider portal, and both carry LOINC codes, units, and the lab’s own reference interval per analyte. A PDF is a dead end: you will spend more time parsing it than analyzing it, and PDF extraction silently loses units and flags.

A workable local schema is one long-format table with columns subject_id, draw_datetime, loinc, analyte, value, unit, method, lab_id, ref_low, ref_high, cva_pct, cvi_pct, fasting_hours, hours_since_last_session. Store it as Parquet, keep the raw HL7 or FHIR payloads immutably alongside, and normalize units at load time rather than at query time. Unit mismatches are the most common corruption in personal lab datasets: ferritin in ng/mL and µg/L are numerically identical, glucose in mg/dL and mmol/L differ by 18.016, and creatinine in mg/dL and µmol/L by 88.4. Also pin the method, because switching from one testosterone immunoassay to LC-MS/MS creates a step change in your series that has nothing to do with you.

With that table in place, the analysis is short. Fit a per-analyte robust location and scale over your own history, flag points exceeding your computed RCV, and join hours_since_last_session and training load so that a CK spike is attributable rather than alarming.

What blood does not tell you

Blood chemistry is a snapshot of circulating concentrations. It does not tell you why a value is where it is. Two athletes with identical ferritin can differ in intestinal iron absorption and in HFE genotype. Two athletes with identical fasting glucose can differ substantially in postprandial excursion, which continuous glucose monitoring shows and a fasting draw cannot. Transcriptomic and proteomic layers add mechanism where a single analyte only gives you level, and research in athletes with repetitive head impact history illustrates how multi-marker profiles separate groups that routine panels do not ⁸.

Two limits worth stating plainly. Blood work of this kind is measurement, not diagnosis. Persistently abnormal hematology, an unexplained high hematocrit, ferritin below your own established range, or any thyroid or hormonal result outside the laboratory interval needs a physician, both for interpretation and because some findings require follow-up you cannot run yourself.

Questions people also ask

Do athletes have high hematocrit? Not consistently. Endurance training expands plasma volume, which dilutes red cells and often lowers measured hematocrit while total hemoglobin mass rises. Hematocrit is a ratio, and its denominator moves with your hydration state and posture at the draw.

What is the fastest way to lower hematocrit? The reversible part is plasma volume: rehydration raises plasma volume and lowers measured hematocrit within hours, which is why a dehydrated draw reads high. Genuine lowering of red cell mass is slow, and pharmacologic effects on hematocrit exist, including the modest lowering seen with renin-angiotensin system inhibition ⁹, but any pharmacologic route is strictly a physician’s decision. A high hematocrit that persists across properly hydrated, properly timed draws needs medical evaluation rather than self-management.

Does exercise reduce hematocrit? Chronic aerobic training tends to reduce measured hematocrit through plasma volume expansion. A single session can transiently raise it through sweat loss and fluid shift. Both effects are about plasma, not red cells, which is why you should compute the plasma volume change between draws before believing a difference.

What does high hematocrit feel like? Often nothing. When symptoms occur, people report headache, flushing, blurred vision, dizziness, or itching. Because it is frequently asymptomatic and because the causes range from dehydration to sleep apnea to primary bone marrow disorders, this belongs in a clinical evaluation and not in a self-interpretation of a home panel.

What blood work is done for teenage athletes? Typically a CBC, ferritin and iron studies, 25(OH)D, a metabolic panel, and thyroid function, with the specific set decided by a pediatrician who knows the athlete’s growth stage and history. Adolescent reference intervals differ from adult ones, and growth alters several analytes, so adult ranges should not be applied.

What blood tests should athletes get? The panel at the top of this page, run on a fixed schedule with fixed preanalytical conditions, is the version we would run: hematology with reticulocytes, full iron studies with hs-CRP, CK, a metabolic and lipid panel with ApoB, 25(OH)D, thyroid, and sex-appropriate hormones. Serial measurement against your own baseline is what makes these numbers mean anything ¹².

Oak builds longitudinal molecular profiles of individuals: whole-genome sequencing, RNA sequencing, proteomics, blood biomarkers, and continuous glucose data, integrated into one model of you. Build your profile.

Footnotes

1. Elaine C. Lee, Maren S. Fragala, Stavros A. Kavouras, et al. Biomarkers in Sports and Exercise: Tracking Health, Performance, and Recovery in Athletes. Journal of Strength and Conditioning Research, 2017. https://doi.org/10.1519/jsc.0000000000002122 ↩ ↩²

2. Robert A. Huggins, Andrea R. Fortunati, Ryan M. Curtis, et al. Monitoring Blood Biomarkers and Training Load Throughout a Collegiate Soccer Season. Journal of Strength and Conditioning Research, 2019. https://doi.org/10.1519/jsc.0000000000002622 ↩ ↩²

3. Anna Nowakowska, Dorota Kostrzewa-Nowak, Rafał Buryta, et al. Blood Biomarkers of Recovery Efficiency in Soccer Players. International Journal of Environmental Research and Public Health, 2019. https://doi.org/10.3390/ijerph16183279 ↩

4. Mario Muñoz-López, Gonzalo Quesada-Fernández, Edgar Simón Sancho-Haro, et al. From Routine Blood Tests to Metabolomics: A Contextual Framework for Interpreting Biomarkers of Training Load, Recovery, and Metabolic Stress in Athletes. Metabolites, 2026. https://doi.org/10.3390/metabo16070483 ↩

5. B.D. Horne, J.B. Muhlestein, D.L. Lappé, et al. Randomized cross-over trial of short-term water-only fasting: Metabolic and cardiovascular consequences. Nutrition, Metabolism and Cardiovascular Diseases, 2013. https://doi.org/10.1016/j.numecd.2012.09.007 ↩

6. Sofie Velghe, Lisa Delahaye, Christophe P. Stove. Is the hematocrit still an issue in quantitative dried blood spot analysis?. Journal of Pharmaceutical and Biomedical Analysis, 2019. https://doi.org/10.1016/j.jpba.2018.10.010 ↩

7. Walter H. Reinhart. The optimum hematocrit. Clinical Hemorheology and Microcirculation, 2016. https://doi.org/10.3233/ch-168032 ↩

8. Alex P. Di Battista, Shawn G. Rhind, Doug Richards, et al. Altered Blood Biomarker Profiles in Athletes with a History of Repetitive Head Impacts. PLOS ONE, 2016. https://doi.org/10.1371/journal.pone.0159929 ↩

9. K. Marathias, B. Agroyannis, T. Mavromoustakos, et al. Hematocrit-lowering Effect Following Inactivation of Renin-Angiotensin System with Angiotensin Converting Enzyme Inhibitors and Angiotensin Receptor Blockers. Current Topics in Medicinal Chemistry, 2004. https://doi.org/10.2174/1568026043451311 ↩