Intervention in the prodromal phase of schizophrenia and related psychoses may result in attenuation, delay or even prevention of the onset of psychosis in some individuals. However, a "prodrome" is difficult to recognise prospectively because of its nonspecific symptoms. This study set out to recruit and follow up subjects at high risk of transition to psychosis with the aim of examining the predictive power for psychosis onset of certain mental state and illness variables.Symptomatic individuals with either a family history of psychotic disorder, schizotypal personality disorder, subthreshold psychotic symptoms or brief transient psychotic symptoms were assessed and followed up monthly for 12 months or until psychosis onset. Twenty of 49 subjects (40.8%) developed a psychotic disorder within 12 months. Some highly significant predictors of psychosis were found: long duration of prodromal symptoms, poor functioning at intake, low-grade psychotic symptoms, depression and disorganization. Combining some predictive variables yielded a strategy for psychosis prediction with good sensitivity (86%), specificity (91%) positive predictive value (80%) and negative predictive value (94%) within 6 months. This study illustrates that it is possible to recruit and follow up individuals at ultra high risk of developing psychosis within a relatively brief follow-up period. Despite low numbers some highly significant predictors of psychosis were found. The findings support the development of more specific preventive strategies targeting the prodromal phase for some individuals at ultra high risk of schizophrenia.