According to Google webmasters:

The simplest robots.txt file uses two rules:
  • User-agent: the robot the following rule applies to
  • Disallow: the URL you want to block
These two lines are considered a single entry in the file. You can include as many entries as you want. You can include multiple Disallow lines and multiple user-agents in one entry.
Each section in the robots.txt file is separate and does not build upon previous sections. For example:
User-agent: *
Disallow: /folder1/

User-Agent: Googlebot
Disallow: /folder2/
In this example only the URLs matching /folder2/ would be disallowed for Googlebot.

User-agents and bots

A user-agent is a specific search engine robot. The Web Robots Database lists many common bots. You can set an entry to apply to a specific bot (by listing the name) or you can set it to apply to all bots (by listing an asterisk). An entry that applies to all bots looks like this:
User-agent: *
Google uses several different bots (user-agents). The bot we use for our web search is Googlebot. Our other bots like Googlebot-Mobile and Googlebot-Image follow rules you set up for Googlebot, but you can set up specific rules for these specific bots as well.

Blocking user-agents

The Disallow line lists the pages you want to block. You can list a specific URL or a pattern. The entry should begin with a forward slash (/).
  • To block the entire site, use a forward slash.
    Disallow: /
  • To block a directory and everything in it, follow the directory name with a forward slash.
    Disallow: /junk-directory/
  • To block a page, list the page.
    Disallow: /private_file.html
  • To remove a specific image from Google Images, add the following:
    User-agent: Googlebot-Image
    Disallow: /images/dogs.jpg 
  • To remove all images on your site from Google Images:
    User-agent: Googlebot-Image
    Disallow: / 
  • To block files of a specific file type (for example, .gif), use the following:
    User-agent: Googlebot
    Disallow: /*.gif$
  • To prevent pages on your site from being crawled, while still displaying AdSense ads on those pages, disallow all bots other than Mediapartners-Google. This keeps the pages from appearing in search results, but allows the Mediapartners-Google robot to analyze the pages to determine the ads to show. The Mediapartners-Google robot doesn't share pages with the other Google user-agents. For example:
    User-agent: *
    Disallow: /
    
    User-agent: Mediapartners-Google
    Allow: /
Note that directives are case-sensitive. For instance, Disallow: /junk_file.asp would block http://www.example.com/junk_file.asp, but would allow http://www.example.com/Junk_file.asp. Googlebot will ignore white-space (in particular empty lines)and unknown directives in the robots.txt.
Googlebot supports submission of Sitemap files through the robots.txt file.

Pattern matching

Googlebot (but not all search engines) respects some pattern matching.
  • To match a sequence of characters, use an asterisk (*). For instance, to block access to all subdirectories that begin with private:
    User-agent: Googlebot
    Disallow: /private*/
  • To block access to all URLs that include a question mark (?) (more specifically, any URL that begins with your domain name, followed by any string, followed by a question mark, followed by any string):
    User-agent: Googlebot
    Disallow: /*?
  • To specify matching the end of a URL, use $. For instance, to block any URLs that end with .xls:
    User-agent: Googlebot 
    Disallow: /*.xls$
    You can use this pattern matching in combination with the Allow directive. For instance, if a ? indicates a session ID, you may want to exclude all URLs that contain them to ensure Googlebot doesn't crawl duplicate pages. But URLs that end with a ? may be the version of the page that you do want included. For this situation, you can set your robots.txt file as follows:
    User-agent: *
    Allow: /*?$
    Disallow: /*?
    The Disallow: / *? directive will block any URL that includes a ? (more specifically, it will block any URL that begins with your domain name, followed by any string, followed by a question mark, followed by any string).
    The Allow: /*?$ directive will allow any URL that ends in a ? (more specifically, it will allow any URL that begins with your domain name, followed by a string, followed by a ?, with no characters after the ?).
Save your robots.txt file by downloading the file or copying the contents to a text file and saving as robots.txt. Save the file to the highest-level directory of your site. The robots.txt file must reside in the root of the domain and must be named "robots.txt". A robots.txt file located in a subdirectory isn't valid, as bots only check for this file in the root of the domain. For instance, http://www.example.com/robots.txt is a valid location, but http://www.example.com/mysite/robots.txt is not.