This paper considers ascertainment corrections for continuous phenotypes. Two main points are considered. The first is a discussion of what ascertainment corrections can be devised to ensure asymptotically unbiased parameter estimates when the nature of the ascertainment procedure is not known. The second is an analysis of the properties of various forms of ascertainment correction possible when the nature of the ascertainment procedure is known. Some ascertainment corrections are thus shown to be valid and others invalid.