Shoots of couch (Elytrigia repens (L.) Desv.) were collected from several grasslands and arable fields from a sand-and a clay area in The Netherlands and grown to adult plants on one experimental field. After three months these plants showed considerable variability which partly appeared to be related to the land use and locality of the sampled fields. However, land use and locality never accounted for more than 10 % of the total variation of the observed plant characteristics. This variability could not be attributed to characteristics of the planted shoots and therefore seemed to be genetically determined and consequently the result of selection of couch types by the environment. The ways in which the land use (grassland versus arable land) might have caused the selection of the couch types are discussed.