Abstract One of the basic concepts in biological fluorescence microscopy is the separation of photons from spectrally distinct fluorophores into different colour channels for studying questions concerning the colocalisation, coclustering or interaction of fluorescently labelled proteins, organelles or membranes on fluorescence images. The quantitative use of colocalisation information, however, is not without pitfalls, and the validity of arguments based on the comparison of dual-colour fluorescence images depends on which molecular targets are labelled and how; which combinations of fluorophores, excitation wavelengths and filters are chosen; how images are acquired and at which spatiotemporal resolution and by which means they are analysed. Unfortunately, no consensual metrics for quantifying colocalisation has emerged. Interrogations as to how colocalisation is best quantified currently make surface again in the context of superresolution microscopies that are increasingly being used in the life sciences. We focus on the practice of the colocalisation imaging workflow and point out problematic steps, guiding the reader to make more informed statements beyond the common ‘red-plus-green-equals-yellow’ approach. Key Concepts Colocalisation measurements are only as good as the raw images they are calculated from – hence, the entire imaging workflow must be understood and quality controlled. Molecules do not occupy the same place. They are, by definition, not ‘colocalised’. Their imperfect images overlap, and both their proximity and the available spatial resolution determine the amount of colocalisation. More often than not, colocalisation will be partial, and therefore only interpretable with some reference or when compared among conditions. Measures of colocalisation need to be held against the random colocalisation arising when structures are placed independently at the same density on an equivalent area. Colocalisation measurements are more meaningful when presented along with positive and negative controls and when the fluctuation of colocalisation within subfields of view, or – for live cells – over time, is reported. Fluorophore concentrations, acquisition parameters, intensity ranges and image processing and segmentation rules should be made explicit to allow comparison of data across experiments and laboratories. Colocalisation studies must always be considered relative to the available spectral, spatial (and for live cells also temporal) resolution, which needs to be explicitly stated.