Attitude measurement has had a somewhat chequered history since the possibility of achieving it successfully was demonstrated by Thurstone in 1929. It has been an important area in science education, particularly in the context of falling uptakes in the physical sciences in many countries, and there have been many attempts to measure learner attitudes to explore why they were deserting studies in such subjects. This paper explores the place of attitude measurement in science education and traces the main approaches that have been developed. The place and nature of attitude scaling techniques is analysed and it is demonstrated that such techniques have many fundamental flaws. These weaknesses make such approaches unlikely to offer the kind of precision needed to take our understanding of attitude development forward in the context of science education. Alternative approaches are outlined and it is strongly suggested that science education research rejects such scaling techniques and moves forward to develop new approaches that can give the kind of detailed analysis which will prove to be positive and useful. This paper seeks to bring evidence from many sources together, to challenge many of the unquestioned assumptions behind the metholodogies used in many attitude-related studies today and to make a positive contribution in encouraging more appropriate methodologies to be adopted more widely. (Contains 2 notes, 1 table and 2 figures.)