Here are a few general findings about TV audiences, confirmed by a number of research studies in different countries.
TV audience sizes in wealthy countries are shrinking, but slowly. When this happens, the audience size may drop slightly two years out of three, but rise slightly in the other year. If you don't draw a long-term graph, you don't see what's happening. And because most advertisers are focused on the present and near future rather than on trends going back a few years, they tend to look at current data and deny there's any change. "What do you mean, our audiences are shrinking? They were up 1% this month!" But maybe down 10% in the last 10 years. This slow shrinkage will probably continue, as more and more alternatives to TV become available.
Most television audience research, particularly audience measurement reports, are sold at very high prices, and you don't find free copies on the Web. However, other types of findings are available, particularly from the ITC in Britain. Since 1998, they have published most of their audience research in PDF (Acrobat) format. You can read this List of published reports from ITC.
It's interesting to look at these reports (e.g. the one on "The Public's View 2002", and reflect on the questions they asked, as well as the questions they chose not to ask. In many ways, the choice of questions tells you more than the survey data does. It seems, for example, that they're very worried about offending people, and that they consider their competition to be other mass media (such as DVDs) but not private media (such as SMS). It's often like this with surveys: the findings tell you more about the organization than about its audience - specially when the questions are framed in such a way that it would be socially undesirable to give certain answers.
The Newsweek Media Research Index also includes some material on TV audiences, but nothing in detail.
Latest findings on TV audiences and sources will be published here soon.