The morphological forms and habits of pharmaceutical crystals are important properties that can be affected by minor changes in operating conditions such as cooling rates and supersaturation. As a result the pharmaceutical industry demands on-line techniques for real-time measurement of the dynamic changes of these properties in crystallisers. On-line imaging represents a potentially powerful technique for real-time monitoring of the morphological forms during crystal growth, but a major challenge is the availability of methods for image analysis that need to be tolerant to the quality of on-line images, accurate, fast and robust. This paper describes a multi-scale segmentation methodology for analysis of images obtained for batch cooling crystallisation of (L)-glutamic acid using an on-line high-speed imaging system developed by the pharmaceutical manufacturer GlaxoSmithKline. The method proves to be able to analyse effectively the on-line images of different crystal morphological forms, and of varied qualities. Application of the methodology to the analysis of images from a different on-line imaging probe and from an off-line slurry sample imaging system demonstrated the capability of generalisation of the method. (C) 2004 Elsevier Ltd. All rights reserved.