This paper, the first in a series devoted to the study of the clinical usefulness of estimations of carcino-embryonic antigen (CEA) in serum and urine, describes in detail a direct radioimmunoassay for CEA in serum and urine (a modified Egan technique, Egan, M.L. et al. (1972), Immunochemistry, 9, 289-299). A comparative study of the behaviour of CEA batches and anti-CEA antisera from different laboratories in the radioimmunoassay is presented. The incidence of increased serum CEA levels in healthy smokers was found to be related to smoking habits. Assays of CEA in serum obtained by the direct technique and the Hansen perchloric acid--zirconyl gel technique (Hansen, H.J. et al. (1971), Clin. Res. 19, 143-147) showed comparable results. Various problems affecting the assay of CEA in serum and urine are critically discussed, e.g. criteria for absorbing-out of anti-CEA antisera, identification of CEA, factors causing false CEA values, use of reference materials, acceptability of a strict cut-off level to indicate increased CEA levels, and factors governing the choice of antisera. In this connection we present a new approach which maintains a reliable and consistent cut-off level in follow-up studies. Evidence is presented that urinary CEA is heterogeneous and perchloric acid-unstable.