By-elections raise fundamental questions, such as: why do third parties manage unexpected breakthroughs in such contests? Why does the government of the day consistently lose support through the `mid-term blues'? Are by-elections essentially idiosyncratic contests reflecting the strengths and weaknesses of individual candidates in particular constituencies? Or can a series of by-election results provide an accurate indication of party popularity in subsequent general elections? Pippa Norris addresses these questions through her analysis of post-war trends in party support. She covers changes in campaigns, contrasting the stable by-elections of the post-war decade with the more volatile ones characteristic of today. She then explores systematic trends in the light of theories of partisan dealignment and retrospective voting, analysing the influence of campaign-specific factors, notably the role of candidates, party organizations, the media, and opinion polls. To set Britain in a comparative context, she also surveys trends in by-elections in Canada and Australia. Finally there is an essential reference section listing changes in party support in almost four hundred British by-elections since 1945.