The AIAG & VDA SPC manual (2026)

The harmonised automotive SPC manual published on 1 July 2026 replaces the 2005 AIAG SPC 2nd edition and the parallel VDA guidance. What changed, what it now expects of your charts, and what that means for tooling.

AIAG & VDA · Our reading of it — the source document is the authority. The document itself ↗

Last reviewed 2026-08-18

On 1 July 2026 the Automotive Industry Action Group and the Verband der Automobilindustrie published the first joint AIAG & VDA SPC manual. It replaces AIAG's SPC 2nd edition — published in 2005, and the reference document for an entire generation of supplier quality engineers — and consolidates the parallel VDA guidance into one text used on both sides of the Atlantic.

If you supply into both regions, the practical headline is that the two glossaries are gone. One set of terms, one set of sampling rules, one set of stability criteria, and no more arguing about whose definition of capability applies to a shared program.

What changed that actually affects your charts

1. Normality is tested, not assumed. The 2005 manual let a great deal of capability analysis proceed on an unexamined assumption of normality. The new one expects the distribution to be assessed first, and provides routes for non-normal data instead of leaving the practitioner to fudge it. If you have been quoting Cpk on flatness, position or runout — all bounded at zero and none of them normal — this is the clause that will come up.

2. New indices for cases the old manual had no answer for. Alongside the familiar Cp/Cpk and Pp/Ppk family:

  • Cw / Cwk describe within-subgroup variation explicitly, separating the question "how tight is the process instant to instant" from "where is it sitting".
  • Pm / Pmk give machine-inherent short-term capability a proper name, which matters for equipment qualification and for low-volume lines where a full process study is not available.

3. Charts beyond Shewhart are in scope. CUSUM and EWMA are treated as legitimate tools for processes with small sustained shifts or time-dependent behaviour, rather than as advanced topics in an appendix. Likewise there is proper treatment of data that is autocorrelated or trending by design — tool wear being the obvious case.

4. Prescriptive where ISO is descriptive. The manual is aligned to the ISO statistical standards (see ISO 7870) but goes further: reporting templates, decision rules, and stated criteria for judging stability. That is a deliberate design choice — it exists to make two suppliers' submissions comparable — and it means less room for interpretation in an audit.

Where to get it

  • The manual is sold by AIAG and through VDA QMC.
  • VDA QMC published a free draft "Yellow Volume" PDF at vda-qmc.de during development. It is a draft, not the published manual, and should be read as background rather than cited as the requirement.

What it means for whatever you use to draw charts

The bar moved in three places at once:

  1. You have to know what your data is before you compute an index on it. A tool that will happily print a Cpk for any column of numbers is now doing you a disservice.
  2. Limits have to be defensible. Which window they came from, when they were set, why they were changed. This has always been good practice; it is now much harder to hand-wave.
  3. Detection speed is a stated concern. If a customer specifies an EWMA or CUSUM scheme for a characteristic, a Shewhart-only tool cannot satisfy it.

Where this platform stands against that

Honestly, part of the way:

  • Shewhart charts to the letter — X̄-R, X̄-s, I-MR, c — with limits from within-subgroup variation and the standard constants, evaluated against the eight Nelson rules with a severity per rule.
  • Limit sets are versioned with their baseline window, author and timestamp, and every change is logged and exportable. That is the "defensible limits" requirement covered.
  • Cp, Cpk, Pp and Ppk side by side, each stated with the sigma it used. The gap between them is reported rather than hidden, which is the same instinct the new manual is acting on.
  • No Cw/Cwk or Pm/Pmk yet. They are on the list; they are not here today.
  • No CUSUM, no EWMA. If a customer mandates one, this is not the tool for that characteristic.
  • No normality testing and no transformations. You must know your distribution yourself; we will not warn you.
  • No MSA, no gauge R&R, no PPAP package. Different manual, different tool.

If you are re-writing your SPC procedure against the new manual, the useful place to start is a list of your special characteristics and which of them are plausibly non-normal. That exercise usually reorganises the project.

Related

Who this applies to

  • SPC for automotive manufacturing — What IATF 16949 and the new AIAG & VDA SPC manual actually ask of a production line, why the capability number on the PPAP is not the point, and where a live chart earns its keep.
  • SPC for aerospace manufacturing — AS9103 asks for the variation of key characteristics to be managed, not just inspected. What that means on a low-volume line, and how to run control charts when a "batch" is eleven parts.

Charts that hold still long enough to be audited

Limits computed from a baseline window and frozen there, every change to them recorded with who made it and when, the rule set stated on the chart, and a read-only view anyone can be sent. That is most of what these documents ask a tool to do.