A historical analysis of the syndrome concept shows that this term has been used in many different ways, ranging from clinical experience to records of coinciding symptoms. However, there seems to be broad agreement on the use of the word ‘syndrome’ in daily practice. If empirical–mathematical methods are applied in syndrome detection, however, a precise operation-alization of the syndrome concept is needed. Traditional procedures have often used models more dictated by methodological considerations than derived from the field of application, i.e. psychiatric syndromatology. An alternative approach, Boolean factor analysis, is presented in this paper. This relatively new method is illustrated by means of the analysis of a small artificial sample with a known structure. As a point of reference, traditional methods (factor analysis, cluster analysis, and multidimensional scaling) are also briefly discussed. It is demonstrated that they all share a deficiency of information about inter-group structure. In contrast, Boolean factor analysis uses a syndromic definition which builds on the basic notion of concurrent symptoms. Moreover, this approach can easily be understood by clinicians.