Over the past 20 years, portable and relatively affordable spectrophotometers have greatly advanced the study of animal coloration. However, the small size of many colour patches poses methodological challenges that have not, to date, been assessed in the literature. Here, we tackle this issue for a reflectance spectrophotometry set-up widely used in ecology and evolution (the beam method). We reviewed the literature on animal coloration reporting the use of reflectance spectrophotometry to explore how the minimum measurable size of a colour patch is determined. We then used coloured plastic sheets to create artificial colour patches, and quantify the relationship between colour patch size and distortions induced by resulting chimeric spectra (spectra contaminated by an adjacent colour patch). Finally, we assessed the generality of our findings using natural colour spots in the lizard Podarcis muralis, as a biologically realistic model. We found a lack of consensus in the literature, frequently resulting in the rejection of valid data or the potential inclusion of unreliable data. As expected, we show that decreasing colour patch size reduces the reliability of reflectance measurements, but also that spectral distortions resulting from chimeric spectra depend on patch/background colour combinations. We found similar results using natural colour spots in P. muralis. We propose a series of steps to avoid the pitfalls described above. First, we provide guidelines on how to identify chimeric spectra and estimate the minimum size of a measurable colour patch in order to avoid them. Second, we show that reducing the probe-to-surface distance allows for more accurate measurements and therefore improves the spectrophotometric assessment of small colour patches. Third, we suggest that, as a general rule of thumb, very small (< 2 mm) colour patches should be avoided when using traditional spectrophotometry methods.