Skip to content

HTTP 403 Forbidden With a Proxy: Causes and Fix Order

Find out who said no, the proxy or the site, then work through the checks in order.

The 3 most common causes

The website refuses this request. 403 is tied to application logic such as missing permissions, a page that needs a login or a role, or a rule of the site. Logging in again does not change it, and a site may even answer 404 to hide a page.

A security rule set by the site owner blocked the request. Cloudflare, for example, returns its "Error 1020 Access denied" page with status 403 when such a rule matches. Which signals a rule looks at is up to the owner, and only the owner can change it.

The proxy refused the tunnel. For an HTTPS page your client first sends CONNECT to the proxy; if the proxy answers 403 there, the request never reached the site. Typical reasons are wrong login or protocol, an expired lease or traffic, or a destination the proxy does not allow.

Which 403 is it?

Which 403 is it?
ItemWhat it meansWhat to do first
curl: CONNECT tunnel failed, response 403The proxy refused to open the tunnel; the site was never contacted.Check protocol and port (HTTP 2222, SOCKS5 3333), the copied login and password, and that the lease or traffic has not run out.
Tunnel opens, then the site answers 403The site refused the request. The proxy did its job.Read the response body, then follow the steps below: account state, region, request rate, IP type.
Cloudflare page "Error 1020 Access denied"A rule of the site owner matched the request.Only the owner can lift it. Send the owner a screenshot with the Ray ID, or check Security Events if the site is yours.
403 with a JSON error such as insufficient permissionsThe key, token or role is not allowed to do this.Fix the permission or the key. A different IP changes nothing here.

Based on MDN (403 Forbidden, CONNECT) and the Cloudflare documentation for Error 1020, read on 16 November 2026. The curl output text was checked against a local test proxy that answers 403 to CONNECT.

Fix order: 7 checks

  1. Step 1: Send the same request without the proxy

    Run the same URL from the same computer with the proxy off. If it also returns 403, the cause is the site, your account or the request itself, not the exit.

  2. Step 2: Find where the 403 comes from

    Run the first curl line below. connect=403 means the proxy refused; connect=200 and status=403 means the site refused.

    Expected result connect=200 status=403 points to the website.

  3. Step 3: Read the response body

    A page or JSON body often says why: a permission, a missing login, a region notice or a security-rule page. Act on that text, not on the number alone.

  4. Step 4: Check the exit IP and country

    Request ipinfo.io through the proxy and compare the country and network with where the page is meant to be used. If the page is limited to certain countries, follow that rule and choose an exit in a supported country.

  5. Step 5: Check the IP type

    A request from a hosting or datacenter range can be refused by sites that expect home connections. See which organization and type the IP shows, and pick the IP type that fits the job.

  6. Step 6: Check your own client

    A script identifies itself with its library name by default (Requests sends a python-requests User-Agent). If the site offers an API, use it with a key: that is what a permission 403 asks for. Follow the site terms for automated access.

  7. Step 7: Stop repeating the request

    An unchanged request is expected to fail the same way, and fast repeats only add traffic. Change one thing at a time, wait a while between tests, and stop when the table says the site is the one refusing.

Quick symptom table

403 on every site, including ipinfo.io

Check Login, protocol and port; whether the lease or traffic has ended in My assets.

Fix Copy the details again from My assets; renew if the term ended.

Stop when If the details are right and the term is active, send the order number to Support.

403 on one site only, others load

Check The response body and whether the page needs a login or is limited by country.

Fix Follow the message: log in with the right account, use a supported country, or use the official API.

Stop when If it is an Error 1020 page, only the site owner can change the rule.

Works in the browser, 403 in your script

Check What the script sends: User-Agent, cookies, the login it uses.

Fix Use the site API with a key, or send what the site documents for automated access.

Find where the 403 comes from

# 1) Where does the 403 come from? connect = proxy answer, status = final answer
curl -sS -o /dev/null -w 'connect=%{http_connect} status=%{http_code}\n' --connect-timeout 10 -x http://LOGIN:PASSWORD@HOST:PORT https://example.com/

# 2) The same request without the proxy, to compare
curl -sS -o /dev/null -w 'status=%{http_code}\n' https://example.com/

# 3) Exit IP and country as the site sees them
curl -sS -x http://LOGIN:PASSWORD@HOST:PORT https://ipinfo.io/json

Example only: replace LOGIN, PASSWORD, HOST and PORT with the values from My assets. connect shows the proxy's answer to CONNECT, status the final answer.

How to do it with Proxy302

Rotating residential proxies cover 195+ countries; the same username and password work on HTTP (port 2222) and SOCKS5 (port 3333) on proxy.proxy302.com. Without a session every new connection gets a new IP; with a session the IP stays fixed. This helps you test whether a 403 depends on the exit, but it does not lift a rule the site set for your account.

For a logged-in account use a static ISP IP: it does not rotate during the lease, and you pick the country on the order page.

Prices and limits

Rotating residential costs US$1.50 per GB on the Budget line and US$5 per GB on the Premium line; purchased traffic has no expiry. A static ISP IP costs US$0.50 for 1 day and US$7.99 for 30 days (7.99 / 30, about US$0.27 a day), the same in every country. There is no free trial; the smallest purchase is 1 GB or one IP for one day.

FAQ

Does a 403 mean my IP is banned?

Not necessarily. 403 only says the server refused the request. It can be a permission, a login, a region rule or a rule of the site owner. Find where it comes from first with the curl line above.

What is the difference between 401, 403 and 407?

401 asks you to authenticate to the site. 403 means authenticating again makes no difference. 407 means the proxy itself wants credentials.

Will a new IP fix a 403?

Only if the refusal depends on the exit. A permission or login 403 stays the same on any IP. Test with and without the proxy before you change anything.

Why do I get 403 in curl but not in the browser?

The two send different things: cookies, a login, headers. Compare what curl sends with -v and use the site API if it has one.

403 on the CONNECT line, what now?

The proxy refused. Check protocol and port, copy the login and password again, and confirm the lease or traffic is active.

Reference links

5
  1. MDN: 403 Forbidden

    Official documentation · Confirmed on:

    Read on 16 November 2026: 403 means the server understood the request but refused to process it; authenticating or re-authenticating makes no difference; the cause is tied to application logic such as insufficient permissions; repeating the request without modification is expected to fail with the same error; a server may send 404 instead to hide a resource.

  2. MDN: CONNECT method

    Official documentation · Confirmed on:

    Read on 16 November 2026: CONNECT asks a proxy to open a TCP tunnel to host:port; any 2XX response means the proxy switches to tunnel mode; the client can send Proxy-Authorization.

  3. Cloudflare docs: Error 1020 Access denied

    Official documentation · Confirmed on:

    Read on 16 November 2026: Error 1020 is returned with HTTP 403 when a security rule set by the website owner blocked the request; a visitor who is not the owner should send the owner a screenshot, and the owner can find the event by Ray ID or IP address in the Security Events log.

  4. Python Requests documentation: Advanced Usage

    Official documentation · Confirmed on:

    Read on 16 November 2026: r.request.headers shows the headers that were sent, including a User-Agent of the form python-requests/x.y.z; proxies are passed with the proxies argument; SOCKS needs requests[socks] and the socks5h scheme resolves DNS on the proxy.

  5. Proxy302 product facts and price list

    Product observation · Confirmed on:

    Product facts file of the site, read on 16 November 2026: rotating residential prices per GB, static ISP lease prices for 1 to 90 days, protocols and ports, countries, no free trial.

Ready to try it?

Create an account, top up once, then pick rotating residential, static ISP or datacenter IPs and have one running in your browser within minutes.