The paper is divided in two parts. The first part compares the data generated by diaries and people meters in Canada. The comparison shows that both ratings systems generate data that is very similar in terms of general viewing behaviour. For example, there is little difference in per capita hours spent watching TV, either in total or by different age/sex and education groups, and there are only small differences in TV usage by day or daypart However, substantial differences were found in the viewing level of cable versus non-cable subscribers and in the audience to various station groups. The second part of the article compares the ratings information produced by the two methods at the program level and introduces the telephone coincidental as a third measure of the audience of a program. While the telephone coincidental tends to reproduce the audience size estimates of the people meter, it confirms the diaries with respect to composition of the audience.