Abstract A group of subclinically depressed students SS ( N = 36) was compared with a group of mildly depressed patients MP ( N = 48) and a group of severely depressed patients SP ( N = 33) on 21 raw symptom severity scores in the Beck Depression Inventory (BDI). Stepwise discriminant function analysis yielded two significant functions, the first representing severity in core symptoms of depression. Symptom severity scores were then adjusted to allow for differences in total BDI scores among the groups and the adjusted scores were reanalysed. Two significant functions were again found, the first representing severity in only three of the core symptoms defining depression according to DSM III. The adjusted scores were further adjusted to equate the groups for differences in score variance and the discriminant analysis was repeated. Again two significant functions were found. The first, clearly representing a dimension of subclinical-clinical depression, was defined by a set of core symptoms smaller than that specified in the DSM III definition of major depression. It is argued that clinical and subclinical depression differ mainly in the absolute severity of the core symptoms defining depression but that subclinical shows a relative emphasis of some, and a relative de-emphasis of other core symptoms. It is concluded that subclinical depression in students offers a fairly accurate ‘model’ of clinical depression and that analogue studies are, to a reasonable extent, justifiable.