Scope Not Holding Zero? How to Tell If It's the Mount, the Ammo, or the Scope

By Optics Intel · July 17, 2026

Your rifle was zeroed. Now it isn’t — the group walked, or the first cold shot landed somewhere new, and every forum thread you’ve found splits evenly between “your rings are loose,” “your scope is broken,” and “you’re flinching.” All three are real possibilities. The way out isn’t opinion; it’s elimination, in the right order — cheapest and most likely first.

First, Confirm the Shift Is Real

Small groups lie. Every rifle and load combination shoots a cone of fire — the honest dispersion it produces even from a perfect rest — and a three-shot group is far too small a sample to locate the center of that cone. Two shots low-left on Saturday doesn’t mean your zero moved; it may mean you finally shot enough rounds to see your real cone.

Before touching a screw:

1. Shoot 8–10 rounds at one aiming point, from a solid rest, with the ammunition you zeroed with.
2. Compare centers, not shots. Find the center of that group and compare it to where your zero used to sit. A genuine shift moves the average; dispersion just decorates the edges.
3. Rule out the obvious rifle-side changes: different ammo lot, suppressor on versus off, a stock or action screw that’s worked loose, a bipod loaded differently than the bags you zeroed from.

If the average point of impact really has moved — and keeps moving or stays wrong on a second confirmation — proceed to the mount.

The Mount Is Guilty Until Proven Innocent

Mounting-system failures cause more “scope won’t hold zero” complaints than scopes do, and they’re free to fix. The mount is a chain with several links, and every link is a candidate:

• Base-to-receiver screws. They live directly against recoil, and they loosen. On many factory rifles they were never properly tight to begin with.
• Ring-to-base interfaces. Cross-slot keys, clamp bolts, dovetails — each one is a joint that can shift.
• Ring caps and the scope tube. Oil or factory preservative on the tube dramatically cuts the rings’ grip; under-torqued caps let the scope slide under recoil, and the shift this produces looks exactly like a scope failure.

The documented record makes the point better than theory. In the Rokslide field evaluations our drop-test database indexes, an SWFA SS 6x42 eval traced an early ~0.2 mil shift to the scope physically slipping in the rings — caps re-torqued from 18 to 25 in-lb, problem gone, and the scope then passed everything including roughly 3,000 rounds of follow-up use. A Zeiss LRP S3 failed drop testing twice in the maker’s factory rings, then passed the full protocol cleanly in different rings — the rings, not the scope, were the failure. And a Burris Fullfield eval ended with the evaluator suspecting the rings rather than the scope, because three different scopes had shifted the same way in that same ring model.

The fix procedure, done once and done right:

1. Degrease the scope tube, ring interiors, and screw threads.
2. Witness-mark every junction with a paint pen: one line across each ring cap seam, ring-to-base joint, and the tube at the ring edge.
3. Torque to the manufacturer’s published values with a real torque wrench — ring caps and base screws have different specs, and hand-feel is how you get one screw at half spec and its neighbor at double.
4. Re-zero, then re-test with another 8–10 round confirmation.

If the witness marks break or the tube creeps relative to the rings, you’ve found your problem and it cost you a marker and an afternoon.

When It Really Is the Scope

Inside every scope, the erector assembly — the tube that holds the reticle or moves the image — is held against spring tension and moved by the turret screws. That mechanism is where zero lives, and it’s where zero dies: impacts can shift it, springs can relax or hang, and turret mechanisms can stick so that dialed adjustments don’t move the point of impact until a later shot or bump jars them loose.

The documented failure signatures are distinctive:

• Failure that survives a mount change. A ZCO ZC420 eval failed its drop testing in the maker’s rings — and failed again a year later in a different ring brand. Once the mount changes and the failure repeats, the mount is eliminated.
• Shifting with every impact. Two separate Leupold VX5-HD samples shifted point of impact with essentially every documented drop, ending several inches from zero.
• Adjustments that don’t take. A VX6-HD eval documented turrets that intermittently failed to move the point of impact during zeroing until further adjustment jarred them — the stiction signature.
• Shift-and-return behavior. A Maven CRS.3 shifted under drops but snapped back to zero at rest — spring-tension behavior that would produce maddening, intermittent wandering in the field. A Trijicon Credo 3-9 shifted about 1.25 MOA after a repeated drop series, then recoil reset it back to zero.

None of this can be diagnosed from the couch, which is the point of the order of operations: by the time you suspect the scope, you should have a 10-round average, intact witness marks, and verified torque — because that evidence is exactly what a warranty department can’t argue with.

The Self-Test, In Order

The full diagnostic, condensed:

1. Confirm: 8–10 rounds, compare group center to old zero.
2. Rifle: action screws, ammo lot, suppressor state, rest technique.
3. Mount: degrease, witness-mark, torque to published spec with a wrench, re-zero, re-confirm.
4. Scope — repeatability checks: dial a large amount up and back (return-to-zero), and a box test if you dial for distance. The documented field evals formalize the stress side of this with drops from 18 and 36 inches with confirmation shots between stages — that protocol, stage by stage, is what the database columns record.
5. Attribute honestly: broken marks = mount. Clean marks plus a repeating shift = scope, and it’s warranty time.

Which Scopes Hold Zero When Someone Actually Checks?

That question now has a structured answer instead of a forum argument: the Scope Drop-Test & Zero-Retention Database indexes every publicly documented evaluation we’ve verified against its source — tracking, 18″ drops, 36″ drops, repeated 36″ series, and long-term round counts, one column per stage, with untested stages marked as exactly that. Some scopes sail through everything including years of documented abuse; some shift and recover; some fail every stage. Most rows are one serial number, so treat them as evidence about tested units, not verdicts on every unit shipped — and read the original write-ups, which every row links.

What Only Range Time Can Answer

This guide is built from documented evaluations and the mechanics of how mounts and erector assemblies work — not from our own destructive testing, and we won’t pretend otherwise. What a structured bench program adds someday: unit-to-unit variance on the same model, before/after measurements with instrumented impacts, and live-fire correlation. When our own bench rows enter the database, they’ll be labeled as exactly what they are.