Change blindness on web pages was studied for 20 participants. The purpose was to find how change blindness appears for web pages, and which changes are easier to detect. The task was to detect if a change had occurred and to show this by the means of the cursor. Rensink's flicker paradigm was used, where four categories of changes were presented. It was easier to detect a change not consisting of a person than one with a person. It was easier to detect a change to the left than to the right. The complexity of the web pages did not appear to have an effect, while large changes were easier to detect than small. The results may indicate that focused attention is differently sensitive for different kinds of changes. They also show that change blindness is a general phenomenon that can be applied to the perception of web pages.