We study the misconceptions about colour that most people hold, determining the general phenomenological laws that govern them. Concept mapping was used to combat the misconceptions which were found in the application of a test specifically designed to determine these misconceptions, while avoiding the possible misleading inductions that could have arisen from the use of everyday language. In particular, care was taken to avoid the distorting effect that the use of the verb "to be" applied to coloured objects could have on the responses. The misconceptions found were shown to have an internal consistency in the form of authentic mini-theories (implicit theories). We compared experimentally the results of two different teaching methods applied to combat these misconceptions. This study was conducted with 470 undergraduates of the University of Extremadura. We analysed the persistence over time of their learning made to overcome those misconceptions. The students were divided randomly into an experimental group (EG) and a control group (CG). To combat their misconceptions, EG were taught following a method based on the use of concept maps, and CG were taught following traditional teaching methods. The results of a pre-test and a post-test were compared for the two groups, finding statistically significant differences. The results allowed the principal working hypothesis to be accepted--concept maps are learning tools which foster conceptual change and allow misconceptions to be eradicated via meaningful learning maintained over time, i.e. EG acquired a relative long-lasting gain in learning that was superior to that acquired by CG. (Contains 6 figures and 6 tables.)