SharePoint over 5,000 items: fix your list view

Content reviewed: 5 November 2026

A request list grows for years without trouble. Then one view stops opening, sorting produces an error and an automation returns suspiciously few records. The settings show only a few thousand items. Before deleting the history, establish which restriction is actually blocking you.

SharePoint Online does not have a storage ceiling of 5,000 items per list. That number refers to the List View Threshold, which protects against expensive database operations. A list can contain up to 30 million items, but that capacity does not guarantee that any view, filter or flow can retrieve everything. SharePoint service limits and the threshold error description address different layers.

This guide applies to SharePoint Online and Microsoft Lists on the same platform. Do not apply SharePoint Server farm settings to it.

Three numbers to keep separate

Reducing the rows per page addresses the third layer. If the query first has to examine an overly broad set to find those rows, the error may remain. Buying extra storage in GB does not change how the list is filtered either.

A note saying 'the list has 12,000 rows' is therefore insufficient for a fix. You need to know which operation the user attempted. Opening the default view, sorting a column and running Power Automate are three separate tests.

An index needs a sufficiently selective filter

An index helps locate records by a column without unnecessarily scanning the whole set. Its existence does not mean that every value in the column returns a small number of records. A Status column can be indexed and still contain tens of thousands of items with the same value.

Microsoft recommends an indexed column that returns fewer items than the threshold as the first determining filter. Further conditions should narrow the result; OR generally broadens it. An indexed lookup column does not solve this threshold issue either. The view-filtering rules are more precise than the general advice to 'index every column'.

Start by measuring the distribution of values. An operational list may divide well by month, department or active process stage. Choose a field that users understand and reliably populate when creating items. An empty or inconsistent key can be technically indexed yet still produce unpredictable operational results.

Model: why a year is too broad but a status can work

Consider a hypothetical list of 60,000 requests. The following counts are inputs to our example, not measurements from a particular organisation.

If a user needs open Sales requests, a sensible design starts with the verified indexed Open filter and then narrows by department. A final result of 900 rows does not retrospectively prove that any filter order or query form would work equally well.

These counts do not provide a universal guarantee of success. Test sorting, grouping and displayed columns in the actual view. The model helps select a promising candidate; its suitability is confirmed by the service's real behaviour.

Data growth can undermine a previously useful filter

Add our own planning assumption: open requests grow by a net 250 per month. After seven months, the filter contains 3,200 + 7 × 250 = 4,950 items. After eight months it contains 5,200. The index is unchanged, but the original design has lost its headroom.

For this model, we would set an internal alert at 4,000 open records and start preparing a further division. The 4,000 trigger is our operational headroom, not another Microsoft limit. Set the review date using your actual growth and seasonality. A December peak in a complaints list may differ markedly from the rest of the year.

That produces a more useful plan than 'archive the list once a year'. You know which filter is approaching trouble, who owns it and which signal should trigger a redesign.

How to repair a specific view

Before changing the shared default view, make a working copy and record the original settings. We recommend this process:

1. Record the exact error, view name, item count and action that triggered the problem.
2. Open the list's indexed-column settings and check the intended filter field. If indexing is still in progress, do not base your test on the assumption that it has finished.
3. Choose a specific value with a demonstrably small matching set and use it as the determining filter.
4. For the baseline test, remove grouping, totals and unnecessary sorting; keep only essential columns.
5. Open the view using an ordinary user's account. Reintroduce other elements one at a time and test after each change.
6. Check several filter values, including the largest group. A small team or the quietest month is not a representative load case.

In its large-list and library guidance, Microsoft recommends simplifying sorting, grouping, totals and column selection when errors occur. Automatically generated indexes on larger lists may also need time to finish in the background.

Do not change filters, permissions and the library structure simultaneously. Five changes at once make it difficult to establish which one was necessary. A working view copy lets you demonstrate the fix while preserving the original settings.

Folders help only when the scope is appropriate

A folder can provide a useful access boundary when people normally work within a particular project or period. A view of all items without folders may not retain that benefit. The documentation also notes that items in the recycle bin can affect the evaluation of some filters. Consider folder and recycle-bin behaviour during diagnosis, rather than treating it as an instruction to delete immediately.

Choose a structure around the business process. If someone needs all open cases across years every morning, mechanically splitting them into annual libraries may make their job harder. Describe the required query before deciding to move data. Do not delete history as a side effect of repairing a view.

Power Automate must demonstrate complete results

A browser view can work while the connected flow misses items. Get items returns 100 items with its default options. Microsoft also describes a case in lists above 5,000 where a filter without the right pagination fails to return matches beyond the initial part of the data. Limit columns by view restricts columns; do not assume it automatically adopts the view's filter. The Get items and Get files documentation distinguishes these settings.

Check the server-side Filter Query, Top Count, pagination and folder scope separately. Increasing the number of retrieved records is not a substitute for a suitable query. A green run status does not establish that every expected item was found.

For the documented Get items issue, open the action's settings and enable Pagination. Choose the requested range according to the actual result the flow must process. Do not try to retrieve the entire large history simply by setting Top Count above 5,000; the documentation limits that value. Then verify the output using control items.

This control set is our editorial test proposal, not a built-in Power Automate feature. Establish the expected count through an independent verified method; using the same faulty flow twice is not an independent check. For a billing export or recurring notifications, retain the test date and filter as well.

What an upgraded licence or admin toggle will not fix

SharePoint Online does not let customers change the List View Threshold. Advice involving Central Administration, Resource Throttling and a daily window for expensive operations applies to SharePoint Server. Upgrading the Microsoft 365 suite is not the usual way to raise this online threshold.

If a storage-full warning accompanies the issue, address it separately. Our article Free up SharePoint storage explains capacity and file versions. GB usage, item counts and the cost of selecting items can change independently.

When choosing the next solution, ask a concrete question: do you need to edit a small operational set, or repeatedly analyse the full history? The latter may call for a different reporting or data design. Do not conclude that migration is necessary merely because a list exceeds five thousand rows.

For an audit, prepare the original view, required queries, counts by key filter and an example of an incomplete flow. Those inputs help establish whether data access can be repaired or the process already needs a structural change.