Before buying a used PWC, or after a misfire, rough idle, or a noticeable loss of power, a compression test answers a question no amount of listening or visual inspection can: how well is each cylinder actually sealing? Rings, valves, and the head gasket all contribute to that seal, and a weak one lets combustion pressure leak past into the crankcase, past a valve, or into an adjacent cylinder or the cooling jacket instead of pushing the piston down. A compression test measures that seal directly rather than inferring it from symptoms.
Running the test correctly
A few steps matter more than people expect, and skipping them produces numbers that don't mean much.
It helps to warm the engine to normal operating temperature first, since cylinder and ring clearances are cut for a hot engine — a cold engine reads artificially low because everything's a little looser than it runs.
It's worth pulling every spark plug, not just the one on the cylinder under test, because cranking against three sealed cylinders and one open one drags the starter down and slows cranking speed, and cranking speed affects how much pressure builds before the gauge's check valve catches it. Pulling all the plugs lets the starter spin the engine at a consistent, faster speed for every cylinder, which is what makes the comparison between them valid.
It's important to disable ignition and fuel delivery per the model's service procedure before cranking — there's no reason for the engine to attempt to fire with the plugs out, and skipping this step invites a fuel or ignition system to do something unpredictable with hands near a spinning flywheel or open cylinder.
Keeping the throttle wide open while cranking matters too: a closed or partially closed throttle restricts the air the engine can pull in, and that restriction shows up in the reading as reduced compression that has nothing to do with how well the cylinder actually seals.
Crank each cylinder through four to six compression pulses — count the gauge needle's kicks — or until the reading stops climbing, and record the peak, repeating the same number of pulses for every cylinder so the comparison stays fair.
Reading the numbers
The figure that matters most isn't any single cylinder's absolute PSI — it's how the cylinders compare to each other. Healthy cylinders on the same engine should read within a similar range of one another; a cylinder reading meaningfully lower than the rest is the one with a sealing problem, regardless of what the absolute number says.
Absolute PSI varies with an engine's compression ratio, and that ratio isn't consistent across a single manufacturer's own lineup, let alone between brands — a naturally aspirated model and a supercharged model built on related architecture routinely run different compression ratios by design, since forced-induction engines are typically built with a lower static compression ratio to leave headroom for boost without inviting detonation. A number that's perfectly healthy on one model can look low next to another model's spec and still be entirely normal. Rather than trusting a PSI figure picked up from a forum post for a given trim level, check that specific model's service manual for its factory spec, and lean on cylinder-to-cylinder consistency as the more reliable read regardless of what the exact target turns out to be.
If a cylinder reads low, a wet test narrows down the cause: add a small amount of oil into that cylinder through the spark plug hole and retest. A reading that jumps up significantly with the oil present points to worn rings — the oil is temporarily filling the gap the rings should be sealing. A reading that barely changes points away from the rings and toward a valve that isn't seating or a head gasket that's breached, since oil in the cylinder doesn't help seal a leak at the valve face or gasket. Two adjacent cylinders both reading low is a particular red flag for head gasket failure between them, and it's worth checking the oil for a milky, coffee-with-cream appearance, which points to coolant mixing with oil through that same kind of breach.
Making sense of dashboard fault codes without a scanner
Compression testing answers questions about mechanical sealing; a lot of day-to-day troubleshooting instead starts with the ski's own gauge throwing a code or a blinking indicator, and every manufacturer handles this differently enough that the specifics are easy to forget between incidents.
The pattern is consistent even where the details aren't: most PWC dashboards can report a fault through the existing gauge — a warning icon, a numeric code on a display, or a blink pattern on an LED where there's no alphanumeric screen to show a code directly. Full diagnostic detail (sensor values, freeze-frame data, fault history) generally requires the manufacturer's dealer-level software, but the basic fault indication on the dash is meant to be readable without one.
What's worth knowing without memorizing a code table: these faults sort into a handful of broad categories that repeat across brands — overheat conditions (often tied to raw water flow or a temperature sensor), low oil pressure or level on four-stroke engines, a sensor circuit fault (throttle position, intake air temperature, crank or cam position), charging or electrical faults, and security or immobilizer issues on skis using a coded key or fob. Many platforms also drop into a reduced-RPM "limp" mode alongside a fault code specifically to limit further damage while the underlying issue gets addressed — reduced power under an active code means the ski is protecting itself, not asking to be run through it.
Because the actual code numbers and blink-count meanings are specific to each manufacturer, and often specific to a model year within that manufacturer, there's no substitute for the fault code table in that model's owner's manual or shop manual. If a blink pattern shows, say, two long flashes followed by three short ones, treat that as an example of the format rather than a code to act on directly — what it actually indicates depends entirely on the platform, and an identical pattern can mean something different from one brand's system to the next. Keeping a copy of that fault code reference on hand, a saved PDF works well, turns a recurring "what did that mean again" moment into a two-minute lookup instead of a guess.
