Nerve Conduction Studies and EMG: What They Actually Measure

The short version#

Nerve conduction studies and EMG are an electrical inspection of your peripheral wiring: the studies tell a clinician where along a nerve the signal is failing and by what mechanism, but they never measure the pain, numbness, or weakness you actually feel. That single fact, that these tests read the wiring and not the sensation, explains almost everything about when they help and when they mislead.

Key points#

• A nerve conduction study measures two things: how long a signal takes to travel (latency) and how big the recorded response is (amplitude).
• Delayed latency suggests damage to the insulating sheath (demyelination); reduced amplitude suggests loss of nerve fibers themselves (axon loss).
• Needle EMG samples muscle. A resting muscle should be electrically silent, and unexpected spontaneous activity signals that muscle has lost its nerve supply.
• These tests are tuned to be very specific, so a positive result is strong confirmation, while a negative result is a weaker rule-out.
• Timing matters: an EMG done in the first two weeks after a nerve injury can look normal before the findings appear.

Two tools, two questions#

Think of an electrodiagnostic study as a two-part inspection carried out in one visit. The nerve conduction part asks whether the cable itself carries current well. The EMG part asks whether the muscle at the end of that cable is still receiving orders. Neither part interprets symptoms; each returns a physical measurement that a clinician then reads against the story you told and the examination you had.

That division of labor is the reason a report can seem to disagree with how you feel. The machine is answering a narrow physiological question, not the broader clinical one.

Nerve conduction: speed and signal size#

In this part, small surface electrodes deliver a brief, mild electrical pulse over a nerve and record what arrives farther along its path. Two numbers carry most of the meaning. Latency is how long the response takes to show up, and a prolonged latency points toward demyelination, injury to the fatty insulation that lets a nerve fire quickly. Amplitude is how large the response is, and a shrunken amplitude points toward axon loss, injury to the conducting fibers themselves. Reading the two together lets an examiner say not just that a nerve is hurt, but roughly how.

EMG: reading the muscle directly#

For the EMG portion, a fine needle electrode is placed into selected muscles and the examiner watches and listens, on a screen and through a speaker, to the electrical chatter. A healthy muscle at rest makes almost no noise. Spontaneous discharges such as fibrillation potentials and positive sharp waves are a sign that muscle fibers have been cut off from their nerve, a state called denervation. When you contract the muscle gently, the size and pattern of the motor unit potentials reveal whether an injury is fresh, old, or already healing through reinnervation.

Timing is a genuine constraint here, and it catches people off guard. Fibrillations rarely surface before roughly two weeks after a nerve injury and may need four to six weeks to fully declare themselves. An EMG ordered too soon can therefore read as reassuringly normal while a real injury is still developing, which is why clinicians often wait several weeks after symptoms begin before scheduling one.

Why these tests are built to confirm, not to screen#

Every diagnostic test lives on a seesaw between two properties. Sensitivity is the share of genuinely affected people a test catches. Specificity is the share of unaffected people it correctly clears. You cannot maximize both at once with a single threshold; loosening the cutoff to catch more early cases invites more false alarms, and tightening it does the opposite.

Electrodiagnostic testing for carpal tunnel is deliberately set toward high specificity, frequently above 95 percent. The logic is that a test whose job is to confirm a suspected diagnosis should almost never flag a healthy nerve, even if that means missing some mild cases. The 2018 discussion in Clinical Neurophysiology Practice puts numbers on the tradeoff: one set of criteria yields roughly 69 percent sensitivity at 97 percent specificity, while another reaches about 92 percent sensitivity but only 63 percent specificity. There is no universally correct cutoff, only the one matched to the question a clinician is asking.

Carpal tunnel: the textbook case, honestly told#

Carpal tunnel syndrome is where these principles are easiest to see. When the median nerve is squeezed at the wrist, its signal slows across that short segment before anything else goes wrong, so a prolonged sensory or motor latency across the wrist is the earliest dependable finding. A 2015 review in Advanced Biomedical Research, drawing on established practice parameters, reports delayed sensory latencies in roughly 49 to 66 percent of affected patients and delayed motor latencies in about 60 to 74 percent, with specificity across these measures running from about 95 to 100 percent. Comparison techniques, which race the median nerve against a neighboring ulnar or radial nerve in the same hand, raise sensitivity further.

Here is the honest limit. The 2018 source draws a distinction that patients rarely hear: the test documents median neuropathy at the wrist, an objective physiological finding, which is not the same as the clinical syndrome a person lives with. The two usually track together, but not always, and the gap runs both ways. Some people with clear symptoms have normal studies, because early or purely intermittent compression can leave conduction measurably intact. And some people with no complaints show abnormal conduction; the same source notes that a meaningful slice of the general population, by some estimates up to a fifth, can have median slowing at the wrist without any symptoms at all. How abnormal the numbers look correlates only modestly with how bad a person feels.

Beyond the wrist#

The same logic travels to other problems. For a suspected radiculopathy, a compressed nerve root in the spine, needle EMG carries modest sensitivity, put by an AANEM practice parameter at roughly 50 to 71 percent for cervical roots, alongside very high specificity. It can confirm and localize an active nerve-root injury but cannot name the exact spinal level with precision, because the muscles it samples draw from overlapping roots. For a generalized polyneuropathy, the study maps the pattern instead: axon versus myelin, sensory versus motor, symmetric versus patchy. That pattern narrows a long list of possible causes without, by itself, naming one.

Across every one of these uses the practical rule holds steady. When the study is positive and clinical suspicion was already high, it is strong confirmation. When the study is negative, it is a softer rule-out, because the physiology it measures can lag behind, or run ahead of, what a person feels. Read next to the history and the examination, it is one of the more objective instruments in neurology. Read on its own, it is a map with the legend torn off.