This totally depends on the task/product and the user-group.
As a web developer, you think that your web-site is the center of the world and that your audience keep track of every change and content update you do. But that's not the case. In most cases, the users don't care. They just want to do what they came to do. Be it registration, buying stuff or finding information.
So I would say that in most cases, it's not a problem.
There is a case study, that documents how a required registration led to a huge dropout. Whereas the redesign to a quick-buy solution increased the sales significantly. I think this is a good example of successful redesign. You should also take look at some of the examples in this question: Case studies to help sell UCD.
That said, some services heavily rely on regular use and regular visits. Social media sites, for instance. Facebook, Google+ and perhaps Youtube depend on users that carry out the same task every day. These kind of sites should be careful with their redesign and only deliver thoroughly tested solutions.
You will always have some users complaining about new design and new features. Sometimes these complaints are justified, but more than often they are just another expression of the reluctance of change. This is an important factor itself, of course, and the user's subjective perception of the system (the satisfaction with the system) should always be taken into the consideration when you're about to deliver something. But if you've done the redesign the right way, then this will be a temporary phase.
Which leads us to the final point. As a vendor, you really need to know that you're doing the right thing. You must have conducted the necessary task- and user-analyzes, and you must have done enough evaluation (user testing, surveys etc) of the redesign. That's the only way to ensure that a redesign will succeed and not turn into a gigantic failure...
