Shock in, wave out: the whole idea
In the previous guide you rebuilt, from the membrane up, how a peripheral nerve carries a signal: a wave of voltage that races along the axon, jumping from gap to gap in the myelin sheath that wraps it. A nerve conduction study (NCS) is simply a way to start that wave on purpose and watch where it ends up. We tape recording electrodes to the skin, deliver a brief electrical pulse over the nerve a known distance away, and the nerve — being an excitable wire — fires. A few milliseconds later, the recording electrodes pick up the arriving signal and the machine draws it as a wave on the screen.
That single wave carries three numbers, and almost everything in this guide hangs on them: how long the signal took to arrive (its delay), how tall it is (its size), and — if you stimulate at two points and do a little arithmetic — how fast it travelled. The art of the study is that these three numbers come apart. A nerve can be slow but full-sized, or fast-arriving but tiny, and those two patterns point to two completely different kinds of injury. A study is not a yes/no test for "nerve damage"; it is a way of asking the nerve what kind of trouble it is in.
Two kinds of wave: motor (CMAP) and sensory (SNAP)
Nerves are mixed bundles, but a study isolates two threads. In a motor study, you stimulate a nerve and record over the muscle it feeds. You are not really recording the nerve here — you are recording the whole muscle's response, summed across all its motor units firing together. That summed response is the compound muscle action potential, or CMAP. Because a muscle is large and its many fibres add up, the CMAP is a big signal, comfortably read through the skin, measured in millivolts.
In a sensory study, you both stimulate and record over the nerve itself, with no muscle in the loop. What returns is the sensory nerve action potential, or SNAP — the pooled response of just the sensory axons. With no muscle to amplify it, the SNAP is a delicate signal a thousand times smaller, measured in microvolts, and easily lost in electrical noise. This fragility is not a nuisance; it is a feature. The SNAP is often the first thing to fade when a nerve is in early trouble, which makes the sensory study one of the most sensitive parts of the whole exam.
The three numbers: latency, amplitude, velocity
Every wave you record is summarised by the trio in latency, amplitude, and conduction velocity. Latency is the clock: the time, in milliseconds, from the moment of the shock to the moment the wave begins. It is mostly a measure of delay — how long the relay took. Amplitude is the height of the wave, from baseline to peak. It is, loosely, a headcount: how much living, conducting tissue actually answered the call. A tall CMAP means many muscle fibres responded; a tall SNAP means many sensory axons did.
Conduction velocity is the speed, in metres per second, and here is the elegant bit: you cannot get it from one shock, because a single latency includes time the signal spent crossing from the last nerve ending into the muscle, which is not pure nerve travel. So for motor studies you stimulate the same nerve at two points, near and far, and record the same muscle both times. Subtract the two latencies and you have cancelled out that shared end-segment; divide the distance between the two shock points by that latency difference, and what remains is the honest speed of the nerve in between. It is the difference, not either number alone, that gives a clean velocity.
MOTOR STUDY, e.g. median nerve to a thumb muscle
stimulate at WRIST ---> record CMAP distal latency = 3.5 ms
stimulate at ELBOW ---> record CMAP prox. latency = 7.5 ms
distance wrist-to-elbow .................. = 240 mm
conduction velocity = distance
------------------------
(prox latency - distal latency)
= 240 mm / (7.5 - 3.5) ms
= 240 mm / 4.0 ms
= 60 m/s <- a healthy forearm speed
Typical normal limb nerve velocities: roughly 40-65 m/sSlow versus small: insulation versus wires
Now the payoff that makes the whole study worth doing. Recall from the membrane guide that speed comes from myelin — the fatty insulation that lets the impulse leap between gaps instead of crawling along bare membrane. So when myelin is damaged, the leap shortens and the signal slows: latency stretches out and velocity drops, but the axons underneath are still alive and still conduct, so the wave, though late, is still reasonably tall. Slowing with preserved amplitude is the signature of demyelination — the insulation has frayed, but the wires are intact.
The other failure is different in kind. If the axons themselves die — the wires are cut, not merely stripped — then fewer fibres remain to answer the shock. The survivors still conduct at a near-normal speed, so latency and velocity can look almost fine; but the wave is short, because there is simply less living tissue to sum. Low amplitude with relatively preserved speed is the signature of axon loss. This is the central reading skill of the whole rung: ask first whether the abnormality lives in the timing (slow = demyelination) or in the height (small = axon loss). The demyelinating-versus-axonal distinction organises almost every diagnosis that follows.
Why care which it is? Because the two carry very different futures, and that shapes the rehab conversation. A stripped axon can re-myelinate fairly quickly and well, so a purely demyelinating block — think of a nerve squashed overnight by an arm draped over a chair — often recovers in weeks. A severed axon must regrow from the injury all the way out to the muscle at a crawling pace of about a millimetre a day, so a hand nerve cut near the elbow may take the better part of a year to reach the fingers, if it gets there at all. That timeline, more than anything, is what a patient with a numb, weak hand is really asking about.
What the study cannot tell you, and what comes next
Honesty about limits is part of reading this test well. A nerve conduction study mostly interrogates the large, fast, myelinated fibres; it is nearly blind to the small unmyelinated fibres that carry pain and temperature, so a patient with burning small-fibre pain can have a perfectly normal study. It samples only a handful of nerves, so a problem in an unsampled nerve is simply missed. And it is a snapshot in time: after an acute axon injury, the nerve segment in the limb keeps conducting normally for several days until the cut-off portion degenerates, so a study done too early can look falsely reassuring. Timing the study matters as much as performing it.
Picture the study in action: a woman who wakes each night with tingling fingers and a hand she has begun to drop things from. We stimulate her median nerve at the wrist and at the elbow, record over a thumb muscle, and find the wrist-to-muscle latency stretched long while the forearm velocity is normal — slowing focused right where the nerve passes under a tight band at the wrist. The wave is delayed but tall: insulation pinched, axons mostly alive. That single pattern, read off three numbers, tells her physiatrist not just the name of the problem but its kind, and so its likely course — and that is precisely the bridge into the next guides, where the needle goes in and these patterns map onto real diagnoses.