Using the 1988 Canadian Election Study I examine why there was only restricted strategic voting in single-member district plurality elections. In that election 19 percent of Canadian voters preferred the party that actually finished third in their constituency, but among these third party supporters only one in eight decided to vote strategically for one of the top two contenders. Strategic voting was relatively rare for two key reasons. First, many third party supporters had a strong preference for their party over all others and were therefore reluctant to rally to either of the top two contenders. Second, many overestimated their party's chance of winning and as a consequence did not feel that their vote would be wasted.In 1951, Maurice Duverger stated what he called a sociological law: 'the simple majority, single ballot system favors the two-party system'. 1 Why should a plurality rule voting method produce a two-party system? Duverger referred to two effects of the plurality rule, the mechanical effect, whereby small parties are systematically underrepresented in Parliament, and the psychological effect, whereby voters do not want to waste their vote on parties that have no chance of winning. The psychological effect induces voters to vote strategically. Cox (1997, p. 71) provides the standard characterization of a strategic (or tactical) voter in a plurality election: 'Some voter, whose favorite candidate has a poor chance of winning, notices that she has a preference between the top two candidates; she then rationally decides to vote for the most preferred of these top two competitors rather than for her overall favorite, because the latter vote has a much smaller chance of actually affecting the outcome than the former'. In a three-way race, then, a strategic voter is someone who votes for her second preference because she perceives that candidate to have better chances of winning than her first preference (Blais and Nadeau, 1996). The existence of strategic voting has been established by a number of empirical studies (for a review, see Alvarez and Nagler (2000)). These studies indicate, however, that there is relatively little strategic voting, typically around 5 percent. 2 In this paper, I seek to understand why relatively few voters cast a strategic vote.The best way here is to focus on those voters who prefer the candidate that finishes third in their constituency. Previous analyses of strategic voting examined the behavior of the entire electorate. But, in theory at least, only supporters of the 'third' candidate in a given constituency have to decide to vote sincerely for their first choice or strategically for their second choice. I use the 1988 Canadian Election Study (CES) because it includes direct questions about voters' perceptions of the various parties' chances of winning in their constituency. 3 As Cox (1997) has shown, these perceptions are central in the decision to cast or not to cast a strate-