Three experiments are described which studied the ability of readers of engineering drawings to solve simple dimension problems as a function of the layout of dimensional information. Different dimensional layouts were compared for (i) different types of object drawn, (ii) different types of dimension problem requiring solution, and (iii) undergraduates and experienced draughtsmen. The results showed that (i) there is an interaction between the type of problem the reader tries to solve and the type of drawn object, (ii) drawing variables additional to the geometry of the object drawn influence a reader's ability to extract dimension information rapidly and correctly, (iii) there are individual differences in performance but all subjects studied in these experiments were affected in a very similar manner by changes in the layout of dimensional information and changes in the type of object drawn, and (iv) there is limited support for a hybrid form of ' abstracted " dimensioning.