A quick release is meant to get slightly looser at the very end of the lever throw. That dip is what locks it shut.

The lever must complete its travel, and overtightening prevents that

This is the whole page. The cam is over-center by design, so clamping force rises through the lever throw, peaks just before the closed position, and eases slightly as the cam passes top dead center. That easing is the lock, because vibration cannot rotate the lever open without first climbing back over the force peak. Sheldon Brown puts the failure directly: if the quick release is overtightened so it does not press down all the way, it will loosen. So the test is not how much force it took, it is whether the lever went fully over. A lever that stops at forty-five degrees because you cranked the nut too far has plenty of clamping force and no lock at all.

Hold the lever and turn the nut. Do not spin the lever

The technique error that produces a wheel which feels secure and is not. Guidance describes it precisely: if the lever is tightened by holding the nut and rotating the lever once the wheel feels tight, the ridges on the lever will have spun rather than biting into place, and the wheel is not properly fitted. The lever is a cam handle, not a wingnut. The correct sequence is to choose where the closed lever will sit, hold it at roughly a hundred and eighty degrees from that final position, tighten the adjusting nut clockwise until you can feel resistance beginning at about halfway through the lever throw, then swing the lever closed. Tension is set by the nut. The lever only closes.

The skewer is not holding the wheel, friction is

Worth understanding because it changes what you check. The axial pull from the cam is only about 0.8 to 1.2 millimeters, and what that pull does is squeeze the dropouts against the locknuts on the hub, so the friction between dropout face and locknut is what actually retains the wheel rather than the skewer itself. Two consequences follow. Those mating faces need to be clean and undamaged, because grease, paint overspray or a burr reduces the friction the whole system depends on. And a skewer that is intact and correctly closed can still be inadequate if the faces it is clamping are contaminated, which is not a failure mode people look for.

Disc brakes changed the math, and there was a recall to prove it

This is why modern practice is stricter than what people remember from rim-brake bikes. Under hard braking, a disc caliper tries to rotate the wheel forward in the fork dropouts, and that load points roughly along the dropout slot, which is exactly the direction the quick release is clamping against. On a rim-brake bike the braking load is symmetric on the rim and does not push the axle out. Documented consequence: recalled Shimano skewers had a lever shape that some riders could close without reaching the over-center lockout, leaving clamping force as low as a thousand newtons, which was adequate for rim brakes and not for a disc rotor pulling the axle out of an open dropout. The fix was a redesigned lever with a clearer closed-position stop plus updated dropout geometry with a retention tab. If your first disc-brake bike is replacing a rim-brake one, this is the habit to update.

Lawyer lips are not an inconvenience, and the rear not having them has a reason

The retention tabs on fork dropouts exist so that merely flipping the cam open does not widen the gap enough for the wheel to clear, which is why removing a front wheel also requires unscrewing the adjusting nut a few turns. That extra step is the safety feature working. Some top-end road bikes omit the lips so a wheel can be changed fast in a race, which is a deliberate trade rather than an oversight. And the front-rear asymmetry has a specific logic, stated plainly in one guide: rear wheels typically have no lawyer lips because the rear wheel cannot come out and throw you off the bike, since all that happens is the wheel jams against the rear stays and the bike skids. The front wheel is the one that ends a ride badly, which is why the tabs are there.

Small details: which side the skewer goes in, and which way the springs face

Two things that get assembled by whatever way the parts fell out of the box. On a disc-brake front wheel, push the skewer through from the drive side so the lever ends up on the opposite side from the rotor and caliper, away from anything hot and anything it can catch on. On a rim-brake bike the side does not matter. The conical springs center the skewer over the dropouts and have an orientation: the narrow end of each spring must rest against the end of the axle. They are not necessary for the wheel to function, which is exactly why they get lost and why a wheel that is finicky to line up is often a wheel missing a spring or running one backwards.

Too much tension reaches the hub bearings

A second reason not to treat maximum force as the goal. The tension in a quick release compresses the hub axle slightly, and that can put the hub bearings out of adjustment. So an overtightened quick release manages to be simultaneously less secure, because the cam never locks, and harder on the hub. There is no version of the story where cranking it further helps. The correct condition is a lever that requires firm hand pressure through the last part of its travel, goes fully over center, and leaves a temporary mark on your palm. That palm mark is the closest thing to a torque wrench a rider has.

A thru-axle is a bolt, and neither technique transfers

Modern bikes increasingly use thru-axles, which thread into the frame or fork and give a much stiffer connection between wheel and frame. That makes them a completely different fastener: a bolt with a torque specification rather than a cam with a feel. Somebody who learned quick releases will under-tighten a thru-axle looking for the over-center snap that does not exist, and somebody who learned thru-axles will crank a quick release like a bolt and defeat the lock. Check which one you have before touching it, and on a thru-axle use the figure printed on the axle or given in the manual rather than a feel, because there is no over-center behavior to tell you when to stop.