UNSCEAR primary reports confirm the world-average dose from atmospheric nuclear testing peaked at 0.11 mSv/yr in 1963 and had declined to about 0.005 mSv/yr by 2008

nuclearradiationtest-ban-treatyunscearquantitative-verificationlow-background-steel

UNSCEAR's 2008 Report to the General Assembly (Volume I), in its summary table of annual radiation doses (Table 1), states that the current worldwide annual average dose from atmospheric nuclear testing is 0.005 mSv, with the note: "The average has fallen from a peak of 0.11 mSv in 1963." The same table lists total natural background as 2.4 mSv/yr; the often-repeated "~4% above natural" framing is arithmetic on those two Tier-1 numbers, not a percentage UNSCEAR states directly.

The figure is not a single report reprinting itself. UNSCEAR's 2000 Report (Annex C, paragraph 61) — an independent assessment cycle published eight years earlier, covering data through the 1990s — states the world average annual dose from atmospheric testing "was highest in 1963 (0.11 mSv) and subsequently declined to less than 0.006 mSv in the 1990s": the same peak value, reached from an earlier vantage point, rather than one number being carried forward unchecked.

This grounds the peak, the 1963 inflection, and the scale of the post-1963 decline claimed in claim-low-background-steel-need-is-fading-as-atmospheric-fallout-decays, which previously carried these same three figures on a Tier-4 Wikipedia citation alone. Checking that article's own footnote shows it cites this same UNSCEAR 2008 report by ISBN — the citation chain, not only the numbers, resolves to this primary. What these figures do not establish is the further inference that low-background steel is therefore "no longer necessary" for sensitive uses — see claim-pandax-ii-2016-modern-steel-still-required-scrap-avoidance-for-low-background for a primary account complicating that leap, and claim-unscear-revised-1963-peak-dose-estimate-0-14-to-0-11-msv for how even the settled 0.11 mSv figure was itself a revision.