Emil Kraepelin proposed to separate psychiatric disorders with psychotic features into two major categories, dementia praecox (later schizophrenia) and manic-depressive insanity (later bipolar disorder and major depression). Over the past century, there have been many efforts to categorize conditions that do not fit readily in either group. These conditions include many cases of acute psychotic illnesses of limited duration, with recovery between recurrences. For some of these conditions, Karl Kleist proposed the term cycloid psychosis: acute features were psychotic, as in schizophrenia, but the course was episodic, as in manic-depression. His concept was later elaborated by Karl Leonhard and Carlo Perris, and validated by modern studies. Leonhard described three overlapping cycloid subtypes (anxiety-beatific, excited-inhibited confusional, and hyperkinetic-akinetic motility dysfunction forms); Perris proposed a more unitary syndrome with operational diagnostic criteria; and recent investigators have considered relatively affective versus thought-disordered subtypes. The cycloid concept is not explicitly included in standard international diagnostic schemes, but both DSM-IV and ICD-10 have broad categories for acute, recurrent psychotic disorders, whose validity remains insecure. We present two cases of probable cycloid psychosis, review the history of the concept, and propose that it be reconsidered as a clinically useful category whose validity and utility for prognosis and treatment can be further tested.