As another poster mentioned, it's a demographics issue. Right now, I'm working on a site used in US K-12 schools, and IE6 usage is around 5%. It doesn't sound like much, but our client can easily justify supporting IE6 based on marginal revenue. We have a large Google Web Toolkit application, so IE6 wrangling is limited to CSS and performance hell -- the JS differences are compiled away.

If you're developing a consumer app, you're probably in pretty good shape though. Some friends at a consumer-focused webshop report IE6 rates < 3%. Furthermore, most of that 3% probably have Firefox installed and could use it easily if they were told that the site wasn't IE6 compatible. So, for them, the ROI of IE6 support is probably quite negative.