The critical importance of marginal constituencies and wards in determining national and local election results in England leads to the question ‘What makes a ward marginal? ’ The results of research on 384 electoral contests over ten years in Birmingham shows that the social class composition of a ward primarily determines its marginality or safety. The class composition associated with a high degree of marginality is as follows: for every non-manual worker there are likely to be twice as many foremen and skilled manual workers, and for every non-manual worker there are likely to be between one and a half and twice as many semi and unskilled manual workers. The further a ward departs from these ratios the more likely it is to be safe for the Labour or Conservative Party. The ratios are highly reliable in picking out marginal and safe wards in normal political periods involving only small swings of electoral opinion between the parties. During abnormal periods, such as the landslide against the Labour Party between 1965 and 1969, the ratios shift slightly but are nevertheless a good indicator of marginal propensity.