Extensive model calculations show that increasing amounts greenhouse gas concentrations in the air cause a cooling of the entire middle and upper atmosphere. Based on a scenario of doubling the CO₂ and CH₄ concentrations, most authors predict a cooling of the order of 10 K in the mesosphere. Observational evidence of cooling is limited, but existing. Temperature measurements by lidar and from satellites show cooling trends. However, these cover only short time intervals, in the order of one or two solar cycles. Evidence in the form of increasing occurence of noctilucent clouds is suggested. Such observations are difficult to quantify reliably in a long-term data set. Lowering of the reflection heights of low frequency radio waves has been presented as evidence of cooling. At thermospheric altitudes Rishbeth predicted a cooling by 50 K to cause lowering of the altitude of maximum electron density in the F2 region by 20 km. Even though various authors report observational evidence for such lowering, others have studied data of many more stations around the world and found a variety of different trends. They conclude that a signal of a greenhouse gas induced cooling of the upper atmosphere cannot be inferred from these data. In this paper, we review the global ionospheric trends and variations observed and compare them with predictions as well as other related data sets, among them data from rockets and from the EISCAT incoherent scatter radars in Northern Scandinavia.