Abstract:The dataset of the Comparative Manifesto Project (CMP) is one the most known and often used datasets in comparative studies of Western Europe. However, the Danish part of the data is very problematic. On average, nearly a third of the content of the Danish manifestos is deemed impossible to code. A close examination of the Danish data reveals large variation in the number of coded sentences across time and parties. An analysis of the manifestos on which the coding is based sets all the warning lights flashing. Firstly, not all manifestos are in fact manifestos; a number consists of other political texts, i.e. speeches, drafts or even stem from different elections, for instance municipal manifestos masquerading as national manifestos. Secondly, the large variation in the length of the manifestos across time suggests that the information derived from them is not readily comparable. This leads to a distorted picture of the positions of the Danish parties. The conclusion of this paper is that in order for the Danish part of the CMP dataset to become more trustworthy it is necessary to search the archives, find the actual manifestos and recode them.