A database consists of tables, each of which has columns and rows. Each row (called a tuple) is a data set that applies to a single item, and each column contains attributes that describe the rows. In database jargon, these columns are called attributes. A database attribute is a column name and the contents of the fields below it in a table.

If you sell products and enter them into a table with columns for ProductName, Price, and ProductID, each of those headings is an attribute. In each field under those headings, you enter the product names, prices, and product IDs, respectively. Each of the field entries is also an attribute. This makes sense, since the non-technical definition of an attribute is that it describes a characteristic or quality of something.

Here is an example from the oft-cited Northwinds database. This database contains tables (also called entities by database designers) for Customers, Employees, and Products, among other things. The Products table defines the attributes of each product. These include a product ID, name, supplier ID (used as a foreign key), quantity, and price. Each of these attributes is an attribute of the table (or entity) called Products.

An attribute is a single piece of data in the tuple to which it belongs. Each tuple is a dataset that applies to one item. The column names are the attributes of a product, and entries in the columns are also attributes of a product.