There has been substantial debate on the measurement of tactical voting in this Journal, much of which has actually been concerned with the definition of a tactical vote. For this reason it is necessary to go ‘back to basics’ to examine the notion of a tactical vote within rational choice theory and follow its implications for the measurement of tactical voting. This Note has four aims: to elucidate the formal theoretical concepts behind the notion of tactical voting; to provide a practical definition of tactical voting based on those concepts; to examine and use the implications of the theory and definition to evaluate existing measures of tactical voting; and to further appraise measures of tactical voting from general methodological considerations. This Note focuses on tactical voting in single-member simple-plurality electoral systems. The following section argues that the range of situations where it may be optimal to vote tactically is both broader and less well defined than previous authors have generally realized. This implies that both the definition, and therefore the measurement, of tactical voting rest mainly on the criteria of voting for a party other than the first choice and doing so in order to best influence who wins. The empirical section examines various approaches to the measurement of tactical voting and argues that the Heath et al. measure is a priori the most attractive available. Furthermore, detailed analysis using British Election Study (BES) data from English voters in 1987, 1992 and 1997, shows that whilst the Heath et al. measure is relatively robust, it can be usefully revised to improve consistency with the strength-of-feeling scores. A corresponding method for identifying a voter's preferred party is also presented.