Some previous studies have investigated the influence of the lighting time course and viewing background on the colour constancy using two-dimensional flat stimuli simulated on a monitor. In the present study, we investigated the categorical colour constancy in real scenes by manipulating (a) the lighting time course, that is, adaptation period to the illuminant (brief adaptation or complete adaptation) and (b) the background structure of a stimulus (a uniform gray background with an approximately 25% spectral reflectance or a multicolour background consisting of the Macbeth ColorChecker and some fruit models). The neutral (u ' = 0.1994, v ' = 0.4671), red (u ' = 0.2433, v ' = 0.4622), green (u ' = 0.1525, v ' = 0.4697), blue (u ' = 0.2049, v ' = 0.4198), and yellow (u ' = 0.1892, v ' = 0.5112) illuminants were produced by an RGB-LED lamp. For each chromatic illumination condition, subjects categorized 240 surfaces with Munsell Value 5/ in four viewing conditions with different combinations of the lighting time course and the background structure. A total of seven subjects participated in experiments with red and green illuminants and five subjects with blue and yellow illuminants. The results showed that the constancy index was the lowest (0.66) in the brief adaptation and gray background condition and the highest (0.74) in the complete adaptation and multicolour background condition. The results suggest that increasing the adaptation period alone or adding chromatic cues in the background with a brief adaptation can help to improve the colour constancy, and a time-taking reference to surrounding coloured objects with the long presentation of the illuminant may also contribute to obtaining colour constancy.