Today internet or websites plays a major role in every person's life. Almost Every sector using the website services like the banking sector using some apps for payment, the passport sector is using for registration and renewal of passport, the government sector for application of PAN cards, adhar cards, etc. This makes human life simpler. But this provides an opportunity to make phishing websites. Phishing websites are the same as legitimate or real websites. The purpose of the phishers is to steal their personal information, account ID, Passwords from individuals and organizations. They even add some tricks by asking security questions like a pet name, city name to gain users' trust. Although legitimate and phishing looks like same there are some features that make difference between that two things. That features such as IP address, URL length, having @ symbol, double slash redirection, Prefix, and suffix, having subdomain, domain registration link, HTTPS tokens, Request URL, URL of anchor, disabling right-click, using a pop-up window, and some more. Already many approaches are proposed for detecting phishing websites machine learning is the most appropriate one. This is because there are some common features that can be identified by machine learning. In this paper, we used a random forest algorithm to detect phishing websites based on features that make difference between both of the websites.