Studies have been carried out in two fully stocked, fast growing Douglas-fir plantations of the Dutch ACIFORN project in three consecutive years, to obtain information on fine root densities (Olsthoorn 1991). For the present paper, data collected in early summer 1987 were used to study the relation of fine root density and proximity to the nearest tree or the dominant tree. A large number of samples (37 in one site and 55 in another) was collected in a small plot (10 x 1 m). Two distances were measured at each sampling point: the distance to the nearest tree and the distance to the tree with a dominant crown above that point. There was large variability in fine root density in the samples. Tests with different regression models showed a distinct rooting pattern for one of the two locations. It is concluded that systematic errors in the assessment of fine root density can arise when sampling points are chosen at a constant distance from trees. For Douglas-fir, this systematic error could have been an overestimation of the fine root density by up to 10%. These systematic errors can be avoided easily, using a stratified random design or a random sampling design. When trees are spaced irregularly, a grid sampling design is also appropriate.