← Back to blog

Datum Shift in GD&T: How It Works and When It Applies

October 23, 2026
Datum Shift in GD&T: How It Works and When It Applies

Datum shift is the movement allowed between a datum feature of size and its fixed simulator when the datum is referenced at maximum material boundary (MMB) or least material boundary (LMB). It is not extra tolerance on the controlled feature. It is freedom for the whole part to move in the datum reference frame, and inspectors who treat it as a bonus accept parts that will not assemble or reject parts that would.

This guide covers when datum shift exists, how to calculate it, how it differs from bonus tolerance, and how to verify it on a CMM without double counting.

QA-Report
Make GD&T Inspection Reports Clear
QA-Report links ballooned drawing dimensions to measured results and flags out-of-tolerance deviations in inspection reports.
Explore QA-Report

TL;DR

  • Datum shift exists only when a datum feature of size (a hole, pin, slot or tab) is referenced with an MMB or LMB modifier. Planar datum features have no size to depart from, so they give no shift.
  • The shift equals the departure of the actual datum feature size from its MMB or LMB limit.
  • Datum shift is not bonus tolerance. Bonus widens the tolerance zone of the controlled feature. Shift moves the whole datum reference frame.
  • At RMB, the default, there is no datum shift.
  • Shift is used once, in one setup, for all features at the same time. You cannot spend it separately on each hole.

Table of Contents

What datum shift means

A datum feature is the physical surface or feature on the part. The datum is the theoretically exact plane, axis or point derived from it. When you apply a datum feature of size at RMB, the simulator adapts to the actual part, so the part is held without play.

When the datum letter carries an MMB or LMB modifier, the simulator is a fixed-size boundary instead. If the actual datum feature is not at that boundary, there is clearance between the part and the simulator. The part can move within that clearance, and that movement is the datum shift.

Comparison of adaptable and fixed datum simulators

The movement is also limited to the degrees of freedom the datum feature constrains. A cylinder used as a secondary datum, for example, can shift in translation across its axis, but it cannot shift along the axis if a primary plane already fixed that direction.

When datum shift applies

Three conditions have to be true at once:

  1. The datum feature is a feature of size.
  2. The datum reference in the feature control frame carries an MMB or LMB modifier.
  3. The actual size of the feature is away from that boundary.

The open study guide on RMB, MMB and LMB notes that, under ASME Y14.5-2009, the datum-compartment modifiers are called RMB, MMB and LMB, to distinguish them from RFS, MMC and LMC applied to the toleranced feature in the tolerance compartment. That wording difference helps you read the frame correctly. A modifier in the datum compartment governs shift. A modifier in the tolerance compartment governs bonus.

The same guide states that at RMB the simulator expands or contracts to the part's mating envelope, so there is no clearance and zero shift.

Why MMB is not the same as MMC

The change in terminology exists for a reason. When a datum feature has its own geometric tolerance, the simulator is not always the size limit of the feature. A GD&T training article on the MMB term gives an example where a 22.0 mm datum hole with a 0.2 perpendicularity tolerance is simulated by a 21.8 gage pin, not 22.0. The boundary includes the effect of the datum feature's own form or orientation control. When you set up the shift calculation, use the boundary stated by the standard, not the raw size limit, if the datum feature is itself controlled.

Datum shift versus bonus tolerance

This confusion is the most common source of bad decisions. A modifier after a datum letter looks like the modifier inside the tolerance compartment, and people treat them as one allowance. They are different mechanisms: bonus tolerance widens the zone around the controlled feature, while datum shift is movement of the frame.

AspectBonus toleranceDatum shift
Where the modifier sitsTolerance compartment (MMC or LMC)Datum compartment (MMB or LMB)
Which feature gets itThe controlled featureThe datum feature of size
What changesTolerance zone sizePosition of the datum reference frame
Can it be spent per featureYes, each feature has its own departure from MMCNo, the frame moves once for the whole part
Effect on the CMM evaluationAdd to the stated toleranceAllow the alignment to float within the clearance

Experienced GD&T instructors describe the mix-up as a widespread industry error. Tec-Ease's tip on datum shift makes the same point: the datum feature gives a shift, not a bonus. Treating shift as extra zone size on every hole in a pattern is how a part passes a report and then fails at assembly.

For the controlled-feature side of the picture, see our guide on bonus tolerance and why no modifier means zero, and the comparison of MMC and LMC and how to choose between them.

How to calculate datum shift

For an internal datum feature at MMB, the shift is the actual size minus the MMB size. The study guide on datum shift behavior gives the same relationship: shift equals the departure of the actual feature size from the MMB or LMB limit.

A worked example

The numbers below are illustrative, not from any specific drawing.

A bracket has a datum B hole of Ø10.00 to Ø10.10 mm. The position callout for a four-hole pattern references datum A, then datum B at MMB.

  • For an internal feature, MMB is the smallest hole, Ø10.00. The simulator is a fixed Ø10.00 pin (assuming datum B has no orientation control of its own).
  • You measure the datum hole at Ø10.06 mm.
  • Shift available: 10.06 minus 10.00 equals 0.06 mm of diametral clearance. The pattern can translate within that clearance relative to the datum B axis.
  • If the hole measured Ø10.10, the clearance would be 0.10 mm. If it measured Ø10.00, there would be no shift at all.

The shift is not added to the Ø0.20 position tolerance of the pattern. It lets the pattern's frame sit in a different place, so the holes can be located relative to a shifted axis instead of the true datum B axis.

For an external feature

For a pin or boss, MMB is the largest size. Shift is the MMB size minus the actual size. A pin that is smaller than its maximum has room to move in the fixed simulator, which is a bushing-like boundary.

LMB

With LMB, the logic flips. The simulator is the least-material size, and shift grows as the feature departs from that limit toward more material. LMB is less common, and it usually appears where wall thickness or edge distance has to be protected.

Why shift is used once, for the whole part

A common mistake is applying datum shift per feature. The frame constrains the entire part, so the shift is a single movement. The datum shift study guide points out that you cannot use leftover shift from one hole to satisfy the position of a hole on the opposite side if the shift directions are mutually exclusive. The constraint must be satisfied in one setup, as it would be on a fixed functional gage.

That is also why the datum shift guide from DatumLab describes the evaluation as a single best-fit analysis in which every feature is checked against its own zone at the same time.

A useful analogy: datum shift is the play between a part and its fixture. You can slide the part left or right in the play, but not both ways for different holes.

A quick check for your own reports

Ask these questions before you accept a position result that relies on shift:

  • Does every feature pass for one common alignment, not a different alignment per feature?
  • Did the shift you used come from the datum feature's measured size, not from the controlled feature's bonus?
  • Is the datum feature of size actually modified at MMB or LMB on the drawing?
  • Does the datum feature itself have a tolerance that changes the boundary?

When to specify MMB or LMB, and when to stay at RMB

RMB gives the strongest repeatability for general metrology because the part is always held at its true mating envelope. That is why it is the default. MMB or LMB makes sense when the part has to assemble with something fixed in size, such as a pin or fastener, and the mating parts have natural play. According to ASME GDTP guidance, MMB and LMB are better where the part meets a fixed-size go-gage at assembly, as they reflect that mechanical play.

For designers, that means choosing the modifier from function. If the mating part is a dowel of fixed size, shift reflects reality, and an RMB callout would reject parts that assemble fine. If the interface is a floating or adjustable one, stay at RMB. For a broader look at how tolerance decisions connect to fixed-size mating parts, an engineer's overview of tolerance in manufacturing is a good companion.

On the inspection side, the modifier changes how you fixture and evaluate the part. Our guides on datum selection for inspection and datum reference frames for metrology teams cover how the frame order and setup affect results.

When to specify MMB or LMB, and when to stay at RMB — overview diagram

Datum shift in CMM inspection

On a CMM, shift is not something you add on a spreadsheet afterward. The software must construct the datum reference frame with the right boundary, then evaluate the controlled features with the frame allowed to float within the clearance. If the CMM program builds the frame as RMB by default, shift is silently removed and a good part can look bad. If a downstream spreadsheet adds the datum clearance to each hole's tolerance, the opposite happens and bad parts can pass.

Common failure points:

  • The program aligns on the measured axis of a datum hole, which is RMB behavior, even though the drawing says MMB.
  • The shift is applied to each feature one at a time.
  • The datum feature's size is not captured, so the shift cannot be calculated.
  • The report shows a single position value with no note on how the datum was modified.

A practical routine:

  1. Read the feature control frame and list every datum modifier.
  2. Record the actual size of each modified datum feature of size.
  3. Confirm in the CMM program how the datum is constructed (RMB, MMB or LMB).
  4. Run one evaluation with all controlled features inside the same alignment.
  5. Put the datum size and boundary in the inspection report so a reviewer can follow the logic.

For related reading, see how datum mistakes cause false rejects, how misaligned datums differ from your fixtures, and our full position tolerance guide.

Our perspective: make the datum logic visible in the report

We see the same pattern again and again in inspection teams. The measurement is fine. The CMM is calibrated. The disagreement comes from how the datum was interpreted, and nobody can tell from the report which interpretation was used. A part is rejected at incoming inspection, the supplier's report says it passed, and both sides are right under different readings of the same callout.

The fix is traceability, not more measurements. A report should tie each ballooned dimension to its measured result, and for modified datums it should show the datum feature, its measured size and the boundary used. If that chain is broken, nobody can tell whether a pass came from a valid shift or from an error.

That is how we built QA-Report. Drawing ballooning, CMM data import and tolerance validation sit in one place, so the callout, the data and the verdict stay connected. For sensitive industries, we also offer on-premise deployment so drawings and measurement data stay under your own control. Our advice to any quality team: write down your rule for datum modifiers once, apply it to every program and review it when a part is disputed.

— Michael Chen

Keep datum logic traceable in your inspection reports

If your team creates FAI, dimensional or GD&T reports, QA-Report links ballooned drawing dimensions to measured results and flags out-of-tolerance values automatically. You can import CMM data into inspection reports or look at our CMM inspection software. Plans on the pricing page include Free at $0 per month, Basic at $49.99 per month and PRO at $149.99 per month, plus on-premise and enterprise options without a published price.

Common mistakes to avoid

  • Adding datum clearance to the position tolerance of each feature.
  • Assuming shift exists on a flat datum surface.
  • Assuming shift exists at RMB.
  • Using the raw size limit as the boundary when the datum feature has its own tolerance.
  • Reporting a position result with no record of datum modifiers.
  • Letting the CMM default to RMB when the drawing calls for MMB.
  • Using different alignments for different features in the same part.

Consequences of misreading are concrete. Tec-Ease notes that confusing shift with bonus can lead to accepting parts that fail to assemble. The opposite error costs money too: rejected good parts, rework and arguments with customers.

Frequently asked questions

Is datum shift the same as bonus tolerance?

No. Bonus tolerance enlarges the tolerance zone of the controlled feature when it departs from MMC or LMC. Datum shift allows the whole datum reference frame to move relative to the part when a datum feature of size is modified at MMB or LMB. Bonus can be calculated per feature, but shift is a single movement shared by all features.

Can a flat surface have datum shift?

No. A planar datum feature has no size, so it has no boundary to depart from and no clearance to consume. Shift needs a feature of size, such as a hole, pin, slot or tab, with an MMB or LMB modifier.

Is there datum shift at RMB?

No. At RMB the simulator adapts to the actual feature, so the part is held without clearance. This is the default in ASME Y14.5-2009 when no modifier is shown, so shift only appears when the drawing explicitly calls for MMB or LMB.

What is the 3-2-1 rule in GD&T?

It is the principle behind a datum reference frame. The primary datum feature needs at least three points of contact, the secondary at least two and the tertiary at least one, so that a rigid part is fully constrained in six degrees of freedom. Datum shift only moves the part in the degrees of freedom left by the datum feature in question.

Sources