The substances in the blood or urine of a pregnant woman which may give an indication of the state of fetal growth are examined. The drawback of measuring such substances is that the values are variable, making it difficult to distinguish between normal and abnormal. Variability arises from technical factors in measurements, from short-term changes of no significance and from the large spread of normal values from one individual to the next. Biochemical parameters of fetal growth can be applied in one of two ways: as screening tests or as control measures by serial assays to guide management. The criteria by which any test should be evaluated--sensitivity, specificity and relative risk--are examined. Particular substances whose measurement may be helpful are considered in terms of the steroids or proteins produced by the fetoplacental unit. The oestrogens, notably oestriol, hold pride of place among the steroids. Dynamic tests of steroid synthesis are also considered. The chief placental proteins of interest are chorionic gonadotrophin, placental lactogen and Schwangerschaftsprotein 1. It is concluded that the method to be recommended is to screen a whole obstetric population with assays of placental lactogen and to follow those with values below the normal limit with serial oestriol assays.