Crystals are particles structured with multiple facets that often have different surface chemistry and hence varied growth rates during crystallisation. It is possible to manipulate the growth of individual facets e.g. via the introduction of tailor-made additives which via changes in the molecular recognition for the different crystal habit faces effect a reduction in the growth rate of a specific face, providing a means for control of the shape as well the size of the final crystalline product. However, the limitations of experimental and analytical techniques for the on-line characterisation of particle shape in real-time have impacted on development and implementation of optimal approaches for design, monitoring and control of crystal shape within the crystalliser. This paper reviews recent developments in on-line shape measurement for crystals grown from solution, provides pointers to future research and development needs and discusses the practical potentials for crystal shape control. Whilst on-line imaging has shown to be the most promising method for on-line measurement of crystal shape within processing vessels, all the work reported in literature has been restricted to 2D imaging, future research and developments should investigate stereo imaging techniques that provide the capability for direct 3D information. In addition, apart from providing shape information, either 2D or 3D, and size, imaging techniques have the potential to provide direct real-time characterisation of the Population balance (PB). Compared with imaging hardware development, image processing has often been the bottleneck for maximizing the use of the direct sensing information, e.g. current image segmentation methods are still restricted to relatively low-solid concentrations, therefore research is needed to develop methods for handling high-solid concentrations. The representation of particle shape using traditional descriptors such as aspect ratio and latent descriptors using principal component analysis and Fourier descriptors are also reviewed. Estimating crystal growth rates and kinetics parameters via image analysis for different crystal facets is still restricted to the simplest case of rod-like crystals that essentially have only two dimensions, the length and width. For the more complicated case of measuring the growth rates of the individual facets, an outline imaging and experimental strategy for direct 3D characterisation is proposed. The paper also reviews recent developments in multi-dimensional PB modelling, i.e., the so-called morphological or polyhedral PB modelling which can provide the evolution of the shape and distributions, of sizes in all crystal face directions in a reactor. Finally, the future perspective for automatic morphology control which requires integration of crystal morphology prediction, morphological PB modelling, on-line 3D imaging and image analysis for shape and PB characterisation as well as computational fluid dynamics is outlined. (C) 2007 Elsevier Ltd. All rights reserved.