Non-ASCII characters in URLs: are they a problem?

What it means

Non-ASCII characters in the URL path are letters, numbers, or symbols outside the basic ASCII set—like Cyrillic, Chinese, or accented Latin characters—appearing directly in the path portion of the URL. Instead of clean ASCII characters, the URL contains Unicode characters that may be served as-is or percent-encoded, depending on how they were generated.

Why it matters

Googlebot crawls and indexes URLs with non-ASCII characters, but inconsistent encoding (mixed percent-encoded and raw Unicode) can cause duplicate URLs to appear in search results. Non-ASCII paths also complicate canonicalization, hreflang resolution, and how AI agents parse and extract structured data from the URL. If the server doesn't properly handle UTF-8 encoding of the path, it can return 4xx errors or serve the wrong resource.

How to fix it

1. Standardize on percent-encoding all non-ASCII characters in URL paths using UTF-8, e.g. /категория/обувь becomes /%D0%BA%D0%B0%D1%82%D0%B5%D0%B3%D0%BE%D1%80%D0%B8%D1%8F/%D0%BE%D0%B1%D1%83%D0%B2%D1%8C.
2. Ensure the server serves UTF-8 content (check <meta charset="utf-8"> and HTTP Content-Type header for non-HTML resources).
3. Verify that the canonical URL, hreflang links, and sitemaps all use the same encoding consistently—never mix percent-encoded and raw Unicode in the same URL.
4. Update internal links to use the canonical form of the URL so crawlers find only one version of each page.

When it's not a problem

Non-ASCII characters are fine if the site is purely regional and the audience expects them, or if the server reliably returns 200 responses with UTF-8 encoding and the URLs are consistent across sitemaps, canonical tags, and internal links. Google has supported UTF-8 URLs for years, so the issue is mainly a hygiene concern rather than a blocking problem.

How common is it?

1187 audited sites in our corpus currently show this issue. The breakdowns below show which platforms, gatekeepers, verticals and countries are most exposed.