Delimitation of the sudden unexplained infant death syndrome (SIDS) is difficult as the diagnosis is made by exclusion. The difficulties in the differential diagnosis are concentrated on interpretation of the significance of positive viral and bacterial findings, inflammatory changes in the respiratory organs, heart and central nervous system together with malformations. Classification of SIDS appears, therefore, to vary according to time and place. New techniques, e.g. DNA analysis, have explained the etiology in a few per cent of the cases but have not yet solved the riddle of SIDS. The article reviews hypotheses about apnoea, arrhythmia, overheating and inefficient surveillance of the infant. It is emphasized that assessment of risk factors for SIDS requires valid epidemiological investigations where the basis for the diagnosis is a uniform classification of SIDS infants as compared with other groups of sudden death in infancy. An investigation of this nature has been initiated in the Nordic countries. It is important to examine and treat infants with abnormal sleep apnoea but generalized employment of monitoring has not reduced the number of unexplained infantile deaths.