This paper considers the relationship between unions and labour productivity in a sample of British engineering firms. Rather than use a single indicator of union presence to determine the union effect a number of measures of unionisation are combined to form a union presence index. An average union non-union productivity differential calculated using this measure is found to be statistically insignificant although there is considerable variation around this average displayed by the firms in the sample. Finn size is also found to be a key determinant of the union impact on productivity and fines with more than 1000 employees are characterised by negative statistically significant union effects. On the other hand, in smaller firms die union impact is neutral although not very well detennined. 77►e results using the index are contrasted with the effects of the closed shop on labour productivity where firm size is also found to be important such that neutral effects occur in small firms but in larger firms the closed shop is associated with lower levels of value added per employee.