In the 20th century, the boundaries of psychosis emerged as an area in which psychiatric judgement faced numerous and profound uncertainties. Between obvious neuroses and personality and reactive disorders on the one hand, and unquestionable psychoses on the other, psychiatrists faced a world of suspected cases of schizophrenia, doubtful personality disorder diagnoses or probable cases of psychosis constituting a garden of equivocal clinical presentations in which both individual psychiatrists and the discipline as a whole were confronted with the limits of (...) their knowledge. This article examines how psychiatrists from two German university clinics managed the multiple uncertainties involved in diagnosing cases of early psychosis between 1950 and 1980. Based on the analysis of a sample of records, we propose a pragmatic interpretation of the ways in which these uncertainties were recorded by psychiatrists. How were uncertainties and doubts expressed in the records and managed by clinicians? What means were used to dispel doubt? What were the consequences for patients of these diagnostic uncertainties? The article defines an uncertainty diagnosis as a diagnosis expressed with reservations by its author and recorded as such in a medical file. Depending on the nature of the uncertainty, the types of evidence used by the professionals and how this evidence was dealt with, we have identified three types of uncertainty diagnoses: suspicion, plausibility and probability diagnoses. The article then reflects on the role of the patients themselves in shaping these uncertain situations. (shrink)