Skip to content

403 Forbidden: What the Status Code Means and How to Respond

When a server answers with 403, it has decided not to act on your request even though the request itself is well formed.

OUTPUT FORMATSHTMLJSONMarkdown

That refusal can come from two very different places: a credential or header the server expected and did not receive, or a server that has judged the traffic to be automated. The checks below help you tell those apart before you change any tooling.

What a 403 Forbidden response actually says

The 403 Forbidden status is a server-side refusal. The request is valid in form, but the server is not willing to fulfil it. That wording matters: a 403 is not a claim that the request was malformed, and it is not a rate-limit or server-error signal. It is a deliberate denial.

Because the denial is deliberate, the server may be enforcing one of two broad policies. It may require credentials or headers the request did not include, or it may have decided the traffic looks automated. Treat those as separate branches rather than one problem.

Branch one: missing or incorrect headers

If the site is normally reachable, start by checking the request's headers. A 403 can come from a missing or incorrect CORS, JWT, or Authentication header that the server expects. Verify each of those against the site's documented requirements before changing anything else.

Use this quick check before escalating:

  • Does the request include the authentication header the endpoint documents?
  • Is the CORS header present where the server expects it?
  • Is the JWT current and correctly formatted?
  • Does the same request succeed with the documented headers and fail without them?

If the answer to the last item is yes, the 403 is a header problem, not an automation problem.

Branch two: the server reads the traffic as automated

When the site is normally accessible and correct headers still do not help, the server may be treating the request as automated. In that case the header checklist will not resolve the 403, because the refusal is about the client's behaviour rather than its credentials. A useful way to test this is to compare the same request made with a documented header set against one made without it: if both are refused, the problem is likely the traffic profile rather than the credentials. If only the header-less request is refused, the header branch is the better explanation and the automation branch can be set aside for now. This comparison is a diagnostic habit, not a guarantee, and it should be repeated before any tooling change is made. When the automation branch does look likely, three practical directions are worth weighing: an undetected browser driver such as undetected-chromedriver, pairing that driver with proxies from a reliable proxy provider, or handing the fetch to a managed web scraping API that absorbs the blocking problem for you. Each direction shifts who owns the traffic pattern, so choose one deliberately rather than stacking them all at once. Before adopting any of them, confirm that the target site's terms and robots directives permit the access you are attempting, and treat a 403 as a signal to respect rather than to force past. If a first-party managed API or proxy service is relevant here, it is covered by the owner's verified offer and proxy-provider details and should only be described once those terms are confirmed. Until then, treat the three directions above as external options to evaluate, not as offers made by this page.

Telling a 403 apart from neighbouring failures

A 403 is often confused with other status codes that also stop a request. The distinction is useful because the remedy differs:

  • 403 Forbidden: the server understood the request and refuses it. Look at headers or automation signals.
  • 429: a rate-limit response. The remedy is usually pacing or backoff, not credentials.
  • 503: a server-side availability problem. The remedy is usually waiting or retrying later.
  • Cloudflare error codes: edge-level blocks with their own documented meanings.

If you are unsure which one you are seeing, read the status line and any error body before assuming a 403.

A worked example of the two branches

Suppose a request to a public page returns 403. You add the documented authentication header and the response becomes 200. That is the header branch: the server wanted credentials it had not received.

Now suppose the same request returns 403 even with the documented headers, while the page loads normally in a browser. That points to the automation branch. The next step is to change the client's traffic profile, not to keep adding headers. Trying more headers in that situation usually produces the same 403 and wastes time.

Choosing a responsible next step

Before changing tooling, confirm that the target site's terms and robots directives permit the access you are attempting. A 403 is a signal from the server, and the appropriate response is to respect it rather than to force past it.

If the access is permitted and the header branch is ruled out, the automation branch offers three practical directions: an undetected browser driver, a reliable proxy provider, or a managed scraping API. The right choice depends on how much of the traffic management you want to own.

Frequently asked questions

What does a 403 status code mean?

It is the Forbidden response status. The server recognises the request as valid but is not willing to fulfil it. The refusal is deliberate, so the fix depends on whether the cause is missing headers or an automation flag.

Should I check headers before changing tools?

Yes. A 403 can be caused by a missing or incorrect CORS, JWT, or Authentication header. Confirm those against the site's documented requirements before assuming the traffic is being flagged as automated.

What if correct headers still return 403?

If the site is normally accessible and the headers are correct, the server may be reading the request as automated. In that case, options include an undetected browser driver, a reliable proxy provider, or a managed web scraping API.

Is a 403 the same as a 429?

No. A 429 is a rate-limit response, while a 403 is a refusal of the request itself. The remedies differ: pacing for 429, and header or automation checks for 403.