Vanguard reported 1,465,147 proxy votes on Form N-PX for the 2025 season. On the 918,179 ballot items the board itself put forward — directors, pay, the auditor, capital — it voted FOR 95.0% of the time. That figure leaves out the 11,441 say-on-pay votes in its institutional-manager report, which by rule covers only executive pay; it voted FOR 95.5% of those. On the 7,776 environmental and social proposals put on the ballot by shareholders, it voted FOR 1.5% of the time.
Ballot items, proposals, companies and funds / accounts include the 12,660 ballot items in Vanguard’s institutional-manager report; the board figure below does not.
On the board’s own proposals
Not included
below: 11,441 say-on-pay votes from Vanguard’s institutional-manager report, which
by rule covers only say-on-pay. It voted FOR 95.5% of them. They are kept apart so the
figure below compares like with like: its funds’ votes on every kind of board proposal.
95.0% FORof 918,179 fund votes on board
proposals in 2024-2025:
directors, pay, auditor, capital.
95.0% FOR
FOR 872,433AGAINST 33,730Abstained or
withheld 12,016
A single “voted with management” figure makes every manager look
the same: director elections are most of the volume and nearly everyone
votes FOR them. The record is therefore reported per category, using the
category the filer itself put on each item in its N-PX.
Vanguard, 2024-2025. “Decided” is FOR + AGAINST, the
denominator of the two percentage columns; abstentions and withheld votes
are shown separately and are never inside a percentage. Counted in ballot
items — one fund’s vote on one proposal at one meeting —
never in shares.
* Fewer than 50 decided items in the row: the FOR and AGAINST counts are shown, not percentages.
An item the filer put in two categories is
counted in both rows, so the rows add up to more than the total. Who
put an item on the ballot is decided item by item: the filer’s own
answer on Form N-PX (Item 1(h)), or, where the filing does not give
one, the proposal’s own wording (13,052
items, shown as inferred). The two figures above use only part of these rows: the
board figure, board proposals on directors, pay, the auditor, capital,
transactions and fund matters; the shareholder figure,
environmental, social, workforce and diversity
shareholder proposals at US-listed companies.
Rows marked “incl. institutional report” include the 12,660 ballot items in
Vanguard’s institutional-manager report; the board figure above does not. How we count these
The list includes the pay votes in Vanguard’s institutional-manager report.
What Vanguard said it would do
The page above is what Vanguard did. Its published proxy voting
policy — what it said it would do — is in the
voting policy library under
Vanguard, with the documents for each year and, where one
exists, a year-on-year comparison.
Funds
Vanguard funds with their own voting-record page, latest season:
The unit is a fund-level observation: one fund’s vote on one
proposal at one meeting. It is not a share, not a filing line and not a proposal.
Vanguard filed 1,465,147 vote records for 2024-2025; after
removing votes filed twice in different N-PX filings, that is
1,163,567 ballot items (fund-level observations) across
12,705 companies and 150 funds. They cover
172,915 distinct proposals: one proposal voted by twenty of its
funds is twenty ballot items here and one proposal. Proposals are counted per
company and meeting, and wordings of the same director nominee are counted
once; two different wordings of any other proposal count twice, so this is an
upper bound. A fund that reported several directions for the same item is a
split vote, counted on its own (see the table below) and not collapsed into the
larger direction. Filers spell the same company several ways;
each company is counted once, its spellings joined by its CUSIP issuer
number.
Share counts are not used anywhere on this page.
Share figures in N-PX carry their own long list of corrections in this
dataset, and a share-weighted rate would inherit every one of them. Every
percentage here is a count of items.
Abstentions are reported, never hidden inside a percentage.
On the board’s own proposals an abstention or a withheld vote is
not support, so it is in the denominator of the FOR figure and not in its
numerator. On shareholder proposals an abstention is neither support nor
opposition, so it is excluded from the denominator and shown on its
own.
11,009 items are outside both headline figures.
Besides the table rows neither figure uses:
Excluded
Items
Why
Not voted
1,250
Reported as not voted, take no action or do not vote. No position was expressed, so there is nothing to count.
Other ballot type
5,406
Frequency ballots (one / two / three years) and anything else that is not a FOR, AGAINST, abstention or withhold.
Direction indeterminate
677
The fund reported several directions for one ballot item with no shares attached to any of them, so which way it voted cannot be established.
Split vote
3,676
The fund voted its shares in more than one direction on the item (for example some FOR and some AGAINST), the smaller part at least 5% of its shares. It is not collapsed into the larger part.
Every figure is derived from Form N-PX, the filing in which registered funds disclose how they voted every proxy and institutional managers disclose their votes on executive pay (say-on-pay, say-on-frequency and golden parachutes).
Figures are
rebuilt weekly; this page was built on November 19, 2026. The
interactive database shows the same votes
company by company, including every individual fund.