What colour tolerances apply to a print run and to brand colours on packaging?

Deviation and variation are separate tolerances

A press run can go wrong in two independent ways: the approved sheet can be off target, or the run can drift away from the approved sheet. Print standards keep the two apart. The deviation tolerance compares the OK sheet with the aim values of the reference printing condition, and describes how well the press was set up. The variation tolerance describes how well it then held. In the offset standard, variation is judged against the OK sheet. Fogra's scheme for digital printing argues for judging it against the run's own average, on at least twenty randomly drawn samples, because a buyer cares about consistency across the delivery, and adds a requirement that at least 70 per cent of the sampled sheets also meet the deviation tolerances.

How the offset tolerances changed, and why nobody can say

The tolerances on the four process solids are the best-known figures in printing, and their history is a caution against treating any tolerance as a constant of nature. The first edition of the offset process standard, in 1996, gave each ink its own ΔE*ab deviation tolerance, ranging from 4 to 8. The 2004 edition replaced them with 5 for all four. The 2013 edition added CIEDE2000 values for information: 3.5 for cyan, magenta and yellow and 5 for black. Researchers who tested these against 185 real jobs reported finding no published justification for any of the magnitudes, and found that the 2013 values pass the four inks at unequal rates. Their analysis suggested black did not need a wider limit than cyan and magenta.

Digital printing: three bands instead of one limit

Digital presses serve work ranging from colour-critical packaging to transactional mail, and Fogra's ProcessStandard Digital answers with three quality types instead of one limit: A, stringent, for colour-critical work; B, normal, for typical industrial printing; C, relaxed. For a side-by-side comparison with the reference, the average ΔE00 over a 72-patch control wedge must be below 2.5, 4.5 or 5.5 respectively, with a 95 per cent quantile below 4.5, 5.5 or 6.5. A second, wider set applies when the comparison is made media-relative, that is, relative to each paper's own white, which is allowed only after a check that the two gamuts are similar in size and shape. The handbook is explicit that the letters describe a level of match, not a better or worse product.

Grey balance is toleranced on its own

A total colour difference treats all errors alike, but a cast in the neutrals is what viewers notice first in a picture, so grey is checked separately in every scheme. The contract-proof standard tests a hue difference on a cyan, magenta and yellow overprint scale. Fogra's bands set a maximum chroma difference on the grey-balance patches, of 2.5, 3.5 or 4.5 for types A, B and C. The G7 method makes neutrality the centre of calibration: its system certification checks tone on black-only and three-colour grey scales as a weighted lightness error and grey balance as a weighted chroma error, each with an average below 1.5 and a maximum below 3. Further G7 levels add colour-difference limits on the solids and on a full characterisation target.

Spot and brand colours

A brand colour printed as a spot ink is judged against its own reference, not against a printing condition. Published figures exist: the contract-proof standard allows 2.5 ΔE00 on spot-colour solids, and Fogra's digital bands set a maximum of 2.5, 3.5 or 5.5, with no media-relative option for spot colours and a recommendation to check first that the chosen press and substrate combination can reach the colour at all. In packaging the brand owner's specification normally governs, and its target increasingly travels as spectral data in the ISO spot-colour exchange format, which is valid only for the ink and substrate that were characterised. Substrate is the usual source of trouble: brightened board, film and foil shift the same ink differently, and a varnish or laminate shifts it again.

The data

What goes wrong, and what to do

What the evidence says

Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.

• StandardModerate evidence

  The 2013 edition of ISO 12647-2 includes informative ΔE00 deviation tolerances for the process solids of 3.5 for cyan, magenta and yellow and 5 for black; the 2004 edition used 5 ΔE*ab for all four solids, and the 1996 edition used a different ΔE*ab value for each.

  Caveat: Taken from a conference paper's account of the three editions; the ΔE00 values are informative, not normative.

  Source: An Investigation of Factors Influencing Color Tolerances; ISO 12647-2:2013 Graphic technology — Process control … — Part 2: Offset lithographic processes

• FactModerate evidence

  A study of 185 jobs in Fogra's Process Standard Offset database found no published justification for the magnitudes of the ISO 12647-2 solid tolerances and found that the 2013 ΔE00 values pass the four inks at unequal rates.

  Measured: 185 offset print jobs in the Fogra PSO database.

  Caveat: One conference paper using one certification database.

  Source: An Investigation of Factors Influencing Color Tolerances

• StandardStrong evidence

  For side-by-side evaluation, Fogra's ProcessStandard Digital requires an average ΔE00 over the MediaWedge below 2.5, 4.5 or 5.5 and a 95 % quantile below 4.5, 5.5 or 6.5 for quality types A, B and C, with a substrate difference below 3.5 for all three.

  Caveat: A Fogra certification scheme for digital printing, communicated as 'PSD 2022'; not an ISO requirement and not applicable to other processes.

  Source: ProcessStandard Digital Handbook 2025: Step by Step toward Printing the Expected

• StandardStrong evidence

  Fogra's ProcessStandard Digital judges variation within a run on at least 20 randomly selected samples against their own average, with a 95 % quantile ΔE00 below 1.5, 3.5 or 5.5 for types A, B and C, and requires at least 70 % of the sampled sheets to meet the deviation tolerances.

  Source: ProcessStandard Digital Handbook 2025: Step by Step toward Printing the Expected

• StandardModerate evidence· North America

  Idealliance's validation process for G7 system certification requires weighted ΔL* on the black-only and three-colour grey scales, and weighted ΔCh on the three-colour scale, each to average below 1.5 with a maximum below 3.

  Caveat: Tolerances for certifying a calibration system, reproduced in a vendor's application data sheet; print-job compliance levels have their own requirements.

  Source: G7 System Certification Application Data Sheet (ChromaChecker 2020.0)

• StandardStrong evidence

  Fogra's ProcessStandard Digital sets a maximum ΔE00 for spot colours of 2.5, 3.5 or 5.5 for quality types A, B and C and provides no media-relative evaluation for spot colours; ISO 12647-7:2016 allows 2.5 ΔE00 on spot-colour solids in contract proofs.

  Source: ProcessStandard Digital Handbook 2025: Step by Step toward Printing the Expected; ISO 12647-7:2016 … — Part 7: Proofing processes working directly from digital data

• FactModerate evidence

  G7 compliance is defined at three levels, G7 Grayscale, G7 Targeted and G7 Colorspace, each adding requirements to the one before: solids and overprints against a reference print condition, then a full characterisation target.

  Caveat: From an instrument maker's summary of the 2018 Idealliance pass/fail document.

  Source: New G7 Pass/Fail Requirements

Related questions

What is the difference between a deviation tolerance and a variation tolerance in printing?

The deviation tolerance limits how far the approved sheet may be from the published aim values; the variation tolerance limits how far sheets in the run may be from a reference, either the approved sheet or the run average. A job can pass one and fail the other.

Why does Fogra's digital scheme have A, B and C levels instead of one tolerance?

Because digital print serves very different needs, from colour-critical packaging to everyday commercial work. The three levels let buyer and printer agree how close a match the job warrants; the handbook stresses that they describe the level of match and are not grades of product quality.

Does passing G7 mean a print job's colours are within tolerance?

Not on its own. The basic G7 level concerns neutral tone and grey balance. Agreement of solids, overprints and a full target with a reference condition belongs to the Targeted and Colorspace levels, and a brand's spot colours are checked separately.

Related topics

Go further

Sources

• An Investigation of Factors Influencing Color Tolerances — TAGA Proceedings 2015, pp. 194 ff. (Lufei Yu, Robert Chung, Bruce Leigh Myers) (Primary research)
• ISO 12647-2:2013 Graphic technology — Process control … — Part 2: Offset lithographic processes — ISO (TC 130) (Official standard or government)
• ProcessStandard Digital Handbook 2025: Step by Step toward Printing the Expected — Fogra Research Institute for Media Technologies (author: Andreas Kraushaar) (Museum or institution)
• G7 System Certification Application Data Sheet (ChromaChecker 2020.0) — Idealliance (Print Properties Committee) (Professional guidance)
• ISO 12647-7:2016 … — Part 7: Proofing processes working directly from digital data — ISO (TC 130) (Official standard or government)
• New G7 Pass/Fail Requirements — Techkon USA (Datacolor) (Manufacturer technical documentation)
• ISO 17972-4:2018 Graphic technology — Colour data exchange format (CxF/X) — Part 4: Spot colour characterisation data (CxF/X-4) — ISO (TC 130) (Official standard or government)
• CxF, AxF, and QTX: Essential File Formats Powering Digital Textile Color Management — X-Rite, Inc. (Manufacturer technical documentation)
• ISO 13655:2017 Graphic technology — Spectral measurement and colorimetric computation for graphic arts images — ISO (TC 130) (Official standard or government)

Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.