Phishing websites is an attempt by an individual or a group to thieve personal confidential information such as the password,credit card information etc from unsuspecting victims for identify theft,the financial gain and other fraudulent activities.Visual cryptography is a special type of secret sharing .in this paper we have proposed a new approach for phishing websites classification to solve the problem of phishing website. Phishing websites comprises a variety of cues within it’s the content-parts as well as the browser-based security indicators provided along with the website. The use of images is explored to preserve the privacy of image captcha by decomposing the original image captcha into two shares that are stored in separate database servers such that the original image captcha can be revealed only when both simultaneously available;the individual sheet images do not reveal the identity of the original image captcha. Once original image captcha is revealed to the used as password.Several solutions have proposed to tackle phishing websites.Nevertheless,there is no single magic bullet that can solve this threat radically.Since anti-phishing solutions aim to predict the website class accurately and that exactly matches the data mining classification.The important features that distinguish phishing websites from legitimate ones and assess how good rulebased data mining classification techniques are in predicting phishing websites and which classification technique is proven to be more reliable.