What fourdc said reminds me of the Canada First Robotics competition. One part was the development of a website, which I did. It had to be in valid xhtml and you had to have graphic and text-only versions of the site. I spent hours making my site determine the users browser and then making sure both sites were laid out exactly the same and that the text only version was still pretty. The winning teams had sites that were incredible for the graphic versions, but sucked ass when it came to the text-only version. What good is my fancy graphical page if the user is browsing with links or lynx.

As for myself. I do everything in Textpad. I'll do a rough hand sketch of what I want the page to look like and i'll pass it off to my roommate. He'll open Photoshop and create the desired image and send it to me. Then i'll draw a grid on it and send it back to him. He'll cut the grid into the pieces while I code the table or frames to hold the images. He'll send them back and i'll drop them in. Then boom our page is graphically designed. Then I usually use an inline frame to display the page content.