Can a Blood Test Show a Viral or Bacterial Infection?

Blood tests can help distinguish between viral and bacterial infections, but no single test does it perfectly. The most widely used markers, including white blood cell counts, C-reactive protein (CRP), and procalcitonin (PCT), all tend to run higher during bacterial infections, and doctors routinely use them to tilt the odds in favor of one diagnosis over the other. The catch is that these markers overlap enough between infection types that interpreting them requires clinical judgment alongside the numbers. Newer approaches, from gene-expression panels to machine-learning algorithms, are closing the accuracy gap, though most haven’t reached everyday clinical use yet.

What a Standard Blood Test Actually Measures

When your doctor orders “bloodwork” to figure out whether you’re fighting a virus or bacteria, the first thing they typically look at is a complete blood count with differential. This breaks down the types of white blood cells circulating in your bloodstream. Bacterial infections tend to drive up the total white blood cell count and, in particular, the proportion of neutrophils (a type of white blood cell that rushes to fight bacteria). Viral infections, by contrast, often push lymphocyte counts higher relative to neutrophils, though that pattern isn’t nearly as reliable as textbooks make it sound.

A study of children hospitalized with respiratory infections found that while white blood cell and neutrophil counts were indeed higher in bacterial cases, the sensitivity at any cutoff was low. In other words, a high count was strong evidence of bacteria, but a normal or low count didn’t rule bacteria out. Specificity, however, was more useful: at a white blood cell cutoff of 20 × 10⁹/L, specificity reached about 95%, and lymphocyte counts turned out to be essentially useless for telling the two apart.¹ That asymmetry is the core frustration with basic blood counts: they can point toward bacteria when values are clearly elevated, but they leave you guessing when results land in the normal range.

CRP and Procalcitonin, the Two Workhorses

Beyond cell counts, two blood-based biomarkers dominate the conversation around sorting viral from bacterial infections: C-reactive protein and procalcitonin. Both are proteins your body produces in response to infection, but they respond to different signals and at different speeds.

CRP is made by the liver and rises in response to inflammation of almost any kind. That’s both its strength and its weakness. It’s cheap, widely available, and fast to measure, but it goes up with autoimmune flares, tissue injury, and even severe viral infections. Values above roughly 80 mg/L tend to suggest a more serious infection, yet CRP alone is considered less accurate than procalcitonin for pinpointing a bacterial cause.² Researchers have acknowledged that while CRP has relevance for differentiating viral from bacterial infection, a reliable threshold remains elusive.³

Procalcitonin is more specific to bacterial infections. Under normal conditions, PCT circulates at very low levels. When bacteria invade, certain immune signals cause many tissues throughout the body to start pumping it out, and levels can rise within a few hours. A systematic review and meta-analysis in febrile infants found that at a cutoff of 0.5 ng/mL, procalcitonin had a pooled sensitivity of about 78% and specificity of about 85% for detecting invasive bacterial infection. CRP at a cutoff of 20 mg/L performed somewhat lower, with a pooled sensitivity near 65% and specificity around 80%.⁴ In a pediatric study comparing procalcitonin levels in children with confirmed bacterial versus viral infections, the sensitivity for diagnosing bacterial infection was about 88% and specificity about 92%.⁵

Neither marker is perfect on its own. In a study comparing PCT, CRP, and white blood cell counts for diagnosing bacterial respiratory infections in children, the individual diagnostic accuracy of each was only moderate when measured by the area under the curve.⁶ This is why doctors almost never rely on a single number. They combine these markers with symptoms, physical exam findings, and other tests to arrive at a diagnosis.

Why Timing Matters

One practical wrinkle that trips people up: these markers don’t all spike at the same moment. Procalcitonin tends to rise within about four to six hours of a bacterial infection taking hold, making it one of the faster signals. CRP is slower, generally peaking one to two days after the infection is well underway.⁷ White blood cell counts can shift within hours but can also be misleadingly normal very early in an illness, or in people whose immune systems are suppressed.

What this means for you is that a blood draw taken in the first few hours of feeling sick may not tell the whole story. If your CRP comes back low at hour three of a fever, that doesn’t guarantee you don’t have a bacterial infection. It may just mean the protein hasn’t had time to accumulate yet. Doctors sometimes repeat labs after 12 to 24 hours specifically to catch that rise.

How Procalcitonin Guides Antibiotic Decisions

Procalcitonin’s real clinical value isn’t just in diagnosis; it’s increasingly used to decide when to start or stop antibiotics. The logic is straightforward: if your procalcitonin is low and stays low, a bacterial infection is unlikely, and antibiotics probably aren’t helping. Across randomized trials and real-world studies, using procalcitonin-guided protocols has shortened antibiotic courses by roughly one to three days in large adult populations, with some smaller studies showing reductions of up to six days. These reductions came without any increase in mortality.⁸

That said, adherence to procalcitonin-based guidelines can be spotty. A hospital study found that meaningful reductions in antibiotic use required adherence rates above 80% to the protocol. When doctors ignored the low procalcitonin results and prescribed antibiotics anyway, the benefit vanished.⁹ And procalcitonin doesn’t always save the day: in patients hospitalized with influenza or RSV, procalcitonin-guided prescribing didn’t significantly change the rate at which antibiotics were started compared to standard care.¹⁰ The evidence is strong for stopping antibiotics sooner, but weaker for preventing unnecessary prescriptions from being written in the first place.

When Procalcitonin Misleads

Procalcitonin is far from foolproof, and it’s worth knowing where it fails. One major pitfall is bacterial superinfection on top of a viral illness. When someone has influenza and then develops a secondary bacterial pneumonia, procalcitonin is sensitive enough to catch that secondary infection in most cases, but its specificity drops. A systematic review of procalcitonin’s performance in influenza patients with suspected bacterial co-infection concluded it works well as a “rule-out” test (a low level makes bacterial co-infection unlikely) but cannot reliably “rule in” the diagnosis on its own because of false positives.¹¹

Another blind spot involves atypical pathogens. Mycoplasma pneumoniae, a common cause of “walking pneumonia,” doesn’t trigger the same inflammatory cascade as typical bacteria. Existing blood signatures designed to broadly distinguish viral from bacterial infections actually perform poorly when mycoplasma is involved, sometimes misclassifying it as a viral infection.¹² That’s a reminder that “bacterial” versus “viral” isn’t always a clean binary.

Beyond Biomarkers: Gene-Expression Tests

The newer frontier is looking not at single proteins in the blood but at how your immune system’s genes respond to different invaders. The concept is that your body mounts measurably different molecular responses to viruses and bacteria, and reading those patterns can reveal which type of infection you’re fighting. These are sometimes called host-response tests.

One such test analyzes the expression of a panel of immune-related genes from a standard blood draw. In a study of over 400 subjects, this approach achieved about 80% accuracy for bacterial infection and about 87% for viral infection in an independent validation group, significantly outperforming procalcitonin’s roughly 69% accuracy in the same cohort.¹³ That’s a meaningful jump. Researchers have also developed a simplified two-gene RNA signature that could be run in under 25 minutes on a portable device, bringing host-response testing closer to something a clinic or emergency department could use at the bedside.¹⁴

An already-available rapid test called FebriDx combines CRP with a second protein, MxA (Myxovirus Resistance Protein A), which is specifically induced by viral infections through the interferon pathway.¹⁵ MxA rises when your body detects a virus and activates interferons, the signaling molecules that coordinate your antiviral defenses.¹⁶ By measuring both a bacterial-leaning marker (CRP) and a viral-leaning marker (MxA), FebriDx demonstrated 87% sensitivity for detecting bacterial infection and 94% specificity for identifying viral infection in emergency department patients.¹⁵ That dual-marker strategy is a meaningful step beyond relying on CRP or procalcitonin alone.

Detecting the Pathogen Directly From Blood

All of the markers discussed so far are indirect. They measure your body’s response to infection, not the pathogen itself. But a separate category of blood tests aims to identify the actual microorganism causing the infection by analyzing fragments of microbial DNA circulating in your bloodstream.

Metagenomic next-generation sequencing (mNGS) takes a sample of blood, extracts all the DNA in it, and uses computational methods to match sequences against databases of known pathogens. A large real-world study found that blood mNGS had a positive detection rate above 80% in surgical and intensive-care patients, with strong consistency compared to standard pathogen cultures. For bacteria and fungi, the test achieved about 89% accuracy and a negative predictive value above 98%, meaning a negative result was highly reliable.¹⁷

A related technology, microbial cell-free DNA (mcfDNA) sequencing, works on a similar principle but specifically targets DNA fragments that pathogens shed into the bloodstream. In children with bone and joint infections, this method identified a pathogen in 81% of initial samples, compared to 68% by traditional culture. It was also positive in two-thirds of cases where cultures came back negative, and it detected pathogens for a longer window of time than blood cultures did.¹⁸ These genomic tests are powerful but still expensive and typically reserved for complex or hard-to-diagnose infections, not routine colds and flus.

Special Considerations for Very Young and Very Old Patients

Age makes a real difference in how useful these tests are. Newborns and very young infants present a particular challenge because their immune systems respond differently than older children and adults, and the stakes of missing a bacterial infection are high. In febrile infants under about two months of age, doctors use risk-stratification criteria that combine blood tests (white cell counts, inflammatory markers, urinalysis) with clinical appearance to sort low-risk from high-risk babies. One validation study of febrile neonates found that these combined criteria had a sensitivity of about 97% and a negative predictive value above 95% for ruling out serious bacterial infection.¹⁹ An earlier and larger evaluation of similar criteria (the Rochester Criteria) reported a negative predictive value of about 99% for serious bacterial infection in well-appearing infants who met all low-risk benchmarks.²⁰ The key insight is that for young infants, no single blood marker works in isolation; it’s the combination of several tests plus a careful clinical exam that makes the system reliable.

At the other end of the age spectrum, elderly patients also present diagnostic challenges. Baseline CRP can be chronically elevated due to age-related inflammation, and immune responses may be blunted, making white cell counts less informative. In older patients, procalcitonin has been studied as a potentially more specific tool. Research in elderly hospitalized patients found that procalcitonin and CRP both rose significantly in those with infection compared to those without, but the optimal cutoffs differed from those used in younger populations.²¹ For sepsis specifically, clinical guidelines emphasize that early diagnosis requires combining bloodwork with a high degree of clinical suspicion, since no single lab value can be trusted as a standalone test.²²

Machine Learning Applied to Routine Blood Tests

One of the more practical advances in recent years involves training algorithms to read patterns across the full panel of routine blood test results rather than relying on any single marker. A machine-learning model built on over 44,000 cases used 16 standard blood values along with CRP, age, and sex to classify viral versus bacterial infections with about 82% accuracy and an area under the curve of 0.905. That outperformed a simple CRP-based decision rule, and the improvement was especially pronounced in the gray zone where CRP alone is ambiguous, between roughly 10 and 40 mg/L.²³

Protein-level signatures are also being explored. A large European collaboration generated proteomic datasets from children with confirmed bacterial and viral infections, using machine learning to identify which blood proteins best distinguish the two groups.²⁴ These approaches are still largely in the research phase, but they signal a shift: the question is no longer whether blood can distinguish the two types of infection, but how well and how quickly.

The Cost Side of Getting It Right

Getting the diagnosis wrong isn’t just a medical problem; it’s expensive. Unnecessary antibiotics for viral infections carry costs from the drugs themselves, from treating side effects like antibiotic-associated diarrhea, and from driving antibiotic resistance in the population. An economic modeling study estimated that widespread use of a rapid point-of-care test like FebriDx for acute respiratory infections in the United States could save roughly $2.5 billion nationally, compared to standard care, by reducing inappropriate antibiotic prescriptions and their downstream consequences.²⁵

For more serious infections like bloodstream infections (sepsis), rapid diagnostic panels that identify the pathogen directly from blood have shown similar economic benefits. One evaluation found that adding a rapid identification panel to standard care saved about $164 per patient in the US while also preventing an estimated 24 deaths per 10,000 patients.²⁶ A separate analysis of point-of-care blood testing for suspected sepsis in Ireland found average savings of roughly €300 per patient, with hundreds of additional lives saved in the modeled population.²⁷ Speed matters: the faster doctors know whether they’re dealing with bacteria or a virus, the sooner they can either start the right antibiotic or skip it entirely.