Proxy Not Working? Why It Fails and How to Fix It (2026 Guide)

By Marcus Reiner · 2026-07-30 · 10 min read · Engineering

Whether your browser says the proxy server is not responding or your scraper suddenly gets 403s, here is how to find the cause and fix it, step by step.

Start here: which problem do you have?

Two very different problems both get called 'proxy not working'. The first is a connection problem on your own device: your browser or app reports that the proxy server is not responding, or shows ERR_PROXY_CONNECTION_FAILED. The second is a blocking problem: the proxy connects fine, but the website returns 403 or 429 errors, captchas or empty pages.

Connection problems are fixed in your settings and credentials. Blocking problems are fixed by changing how your requests look and which kind of IP you use. Pick the section that matches what you see.

Fix 'proxy server isn't responding' on Windows, Chrome and Mac

Chrome and Edge use your operating system's proxy settings, so the fix is usually there. On Windows 11, open Settings, then Network and internet, then Proxy, and turn off 'Use a proxy server' under manual setup. Turn on 'Automatically detect settings' if you are not meant to use a proxy at all. In Chrome you can reach the same screen from Settings, System, 'Open your computer's proxy settings'.

On a Mac, open System Settings, then Network, choose your connection, click Details, open Proxies and untick any protocol that is enabled. If you do use a proxy, check that the host and port are correct and that the proxy is online.

If the error persists, disable browser extensions and any VPN or security software that sets its own proxy. On Windows you can also open Command Prompt as administrator and run netsh winhttp reset proxy, then ipconfig /flushdns, and restart the browser.

Check the basics before blaming the provider

Most proxy failures come from setup mistakes. Confirm the host, port, username and password exactly as shown in your provider dashboard. Make sure you chose the right protocol: an HTTP proxy will not work if your tool is set to SOCKS5, and the reverse.

If your plan uses IP whitelisting instead of a password, your own IP may have changed. Re-add it in the dashboard. Also check that your plan has not run out of bandwidth, expired, or hit a thread limit.

A quick test isolates the problem: run curl with the proxy flag against a simple page, for example curl -x http://user:pass@host:port https://httpbin.org/ip. If that returns a proxy IP, the proxy works and the problem is in your tool or target site. If it fails, the problem is the proxy settings or the account.

What proxy error codes mean

407 Proxy Authentication Required: the proxy rejected your credentials or your IP is not whitelisted. Fix the username, password or whitelist entry.

403 Forbidden: the target site understood the request and refused it. Your IP, headers or fingerprint were flagged. This is a blocking problem, not a connection problem.

429 Too Many Requests: you are sending requests too fast from one IP. Slow down, add delays, or rotate IPs.

502, 503 and 504: the proxy or the site's gateway failed or timed out. Retry with backoff, try a different location, and check the provider's status page.

ERR_TUNNEL_CONNECTION_FAILED and ERR_PROXY_CONNECTION_FAILED: the browser could not reach or build a tunnel through the proxy. Check host, port, protocol, credentials and whether the proxy is online.

Why scrapers get blocked even when the proxy works

A working proxy only changes your IP address. Websites also look at the IP type, how fast you send requests, your headers, your TLS and browser fingerprint, whether JavaScript runs, and whether your behavior looks human. Failing any of these can get you blocked.

The common causes are: using datacenter IPs on a site that blocks them, sending too many requests per IP, a default HTTP library fingerprint that does not look like a real browser, missing or inconsistent headers, rotating the IP in the middle of a session that expects one identity, and a location that does not match the language or timezone your requests claim.

Fixes follow from the cause: switch to residential or ISP IPs, lower the rate per IP, use a real browser or a library with browser-like fingerprints, keep headers consistent, and use sticky sessions when a login or cart is involved.

Why it worked yesterday and fails today

Sites update their detection continually, and the IPs you were using may have been flagged by other users of the same pool. A setup that is close to the line can fall over with no change on your side. Some platforms have also published rules about automated access, so check the current terms for any site you depend on.

Build in monitoring: track your success rate by target and by IP type, and alert when it drops. That turns a mystery failure into a quick diagnosis.

Which proxy type to switch to

Datacenter proxies are cheap and fast, and they are the first to be blocked on heavily protected sites. Residential proxies use IPs assigned to home connections and are blocked less often. ISP (static residential) proxies combine a stable address with residential reputation, which suits account work. Mobile proxies are the hardest to block and the most expensive.

Datacenter proxies are not useless. They still work well on lightly protected targets and for high-volume, low-risk tasks. The mistake is using them on sites that explicitly detect them.

Check how your provider sources its IPs

Residential IPs should come from users who knowingly opted in. Reporting this year linked some proxy networks to devices that were enrolled without clear consent, and brands tied to the IPIDEA network were disrupted in early 2026. Providers can lose access to IPs or shut down, which breaks setups built on them.

Ask any provider how its residential IPs are sourced. A vague answer or signup with no verification at all is a warning sign. Spreading critical work across more than one provider also reduces the damage if one fails.

A note on legality

In the US, courts have treated accessing public web pages differently from breaking into protected systems, and the Computer Fraud and Abuse Act has been interpreted narrowly in cases such as hiQ v. LinkedIn and Van Buren. Separate from that, most sites' terms of service prohibit automated access, which can lead to account bans or civil claims. Rules also differ by country and by data type, especially personal data. This is general information, not legal advice, so talk to a lawyer about anything high-stakes.

Quick troubleshooting checklist

1. Can the proxy fetch a simple page with curl? If not, fix credentials, host, port, protocol or whitelist. 2. Does the error say 407? Fix authentication. 3. Does it say 403 or 429? Slow down and change IP type or fingerprint. 4. Does it work on one site and not another? The second site blocks your IP type. 5. Did it break after working? Check plan limits, then your success rate by IP pool, then the target site's recent changes.

Frequently Asked Questions