We have counted the number of references in 1179 papers published in Astronomy and Astrophysics over twenty years. The number of references has increased by 60% between 1975 and 1995, reflecting the increase (by the same amount) of the literature which must be cited, and of the number of pages per paper. There are 1.5 times more references in predominantly observational fields than in others. References are used 1.65 times in the text, and there is no trend with time or field. They appear mostly in the introduction (30%) and in the main body of the paper (60%), but papers in predominantly observational fields tend to use less references in the introduction and more in the main body than papers in the other fields. Most references (62%) serve to support a result, and tend to be of theoretical nature. Astronomers are a very conformist bunch, as there are no trends with nationality, and references to conflicting evidence are kept at about 8%. The analysis of a series of papers by de Vaucouleurs on the Hubble constant shows how a controversial subject affects the use of references.