According to the dividend signalling hypothesis, dividend change announcements trigger share returns because they convey information about management's assessment on firms' future prospects. We analyse the classical assumptions of the dividend signalling hypothesis, using data from three European countries. The evidence gives no support to a positive relation between dividend change announcements and the market reaction for French firms, and only weak support for the Portuguese and UK firms. After accounting for non-linearity in the mean reversion process, the global results do not give support to the assumption that dividend change announcements are positively related with future earnings changes. We also formulate two hypotheses in order to explore the window dressing phenomenon and the maturity hypothesis, finding some evidence in favour of both, especially in the UK market.