In general, there should be a set of six to eight (6 – 8) tabs from which the user can select. Using more than eight (8) tabs can significantly reduce usability, as you are asking the user to select a single item from a domain set that is too large.
Typically, the information to be shown on the tabs should be categorized in a way that matches the user’s conceptual model of the data. The goal of tabs is to instantly show the user the available information and make it reachable with one click. The information on each tab is typically unrelated to the information on the other tabs, and, ideally, the information is already known to the user or is easy to categorize. While tabs can be categories, they can also be points on a discrete scale; for example, the alphabet in an alphabetical index.
Tabs should be presented like tabular sheets. This relationship can be enforced by showing the tabular sheets visually, i.e., connecting the information area to the current tab.
The tabs should also be presented horizontally. This places an important limitation on the number of tabs that fit in the layout. This number is usually smaller than ten (10), unless an alphabetical index is being used.
The purpose of the tab pattern is to allow the user to easily see all of the available options and select one. When implemented correctly, the user can easily relate the tab metaphor to the tabs on a traditional set of paper files. Easy recognition of the selected tab and a basic understanding of how to make a selection should come naturally to the user. The tab pattern can also be used to show a complex object as whole, with each of its peer components displayed as a separate tab.