This research reports on the colour evolution of six rosé wines during sixteen months of storage in the bottle. Colour changes were determined in terms of CIELAB colour parameters and in terms of the common colour categories used in visual assessment. The colour measurement method reproduces the visual assessment conditions during wine tasting with respect to wine sampler, illuminating source, observing background and sample-observer geometry. CIELAB L*, a*, b*, C* and h ab colour coordinates were determined at seven different times (t = 0, 20, 80, 153, 217, 300 and 473 days). The time evolution of colour coordinate values was studied using models related to linear, quadratic and exponential rise to a maximum. Adjusted R 2 , average standard error and CIELAB ΔE* colour difference were used to compare models and evaluate their performance. For each colour coordinate, the accuracy of model predictions was similar to the standard deviation associated with a single measurement. An average ΔE* = 0.92 with a 90 percentile value ΔE* 90% = 1.50 was obtained between measured and predicted colour. These values are smaller than human colour discrimination thresholds. The classification into colour categories at different times depends on the wine sample. It was found that all wines take three to four months to change from raspberry to strawberry colour and seven to eight months to reach the redcurrant category. Only two wines had developed salmon colour by the end of the experiment.