AbstractWe present a practical solution to the problem of reliably relocating a cell or similar microscopic object, in the presence of inaccurate stage mechanics. Our technique involves storing the relative coordinates of a ‘constellation’ of nearby slide objects, which is then used in a Bayesian statistical model to select the most likely object upon subsequent relocation. This approach has the significant benefit of assigning honest posterior probabilities, so enabling uncertain relocations to be identified. Results are presented for a cervical cytology application. Experimentally derived calibration curves are given to argue the statistical reliability of the mo