In the general population, symptoms of mania and psychosis are more broadly distributed than their associated clinical syndromes. Little is known, however, about how these subclinical population phenotypes co-vary with and impact on each other.In a representative population cohort of 7076 adults, prevalence of mania and psychosis symptoms and syndromes were assessed with the CIDI at baseline, at one (T1) and two years later (T2). The degree of comorbidity between subclinical mania and subclinical psychosis was examined, as well as the impact of subclinical comorbidity on social impairment and transition from subclinical mania to onset of bipolar disorder.The lifetime prevalence of at least one manic and one psychotic symptom was 4.1% and 4.2% respectively. Excluding individuals with any lifetime DSM-III-R bipolar or psychotic disorder (n=218), these prevalences were 2.3% (subclinical mania) and 2.8% (subclinical psychosis). Individuals with subclinical mania had a 17% risk of subclinical psychosis, compared with 2.3% in those without (P<0.000). Comorbid subclinical psychosis in individuals with subclinical mania was much more predictive of a future diagnosis of bipolar disorder (positive predictive values of 3% versus 10% respectively).Subclinical phenotypes of mania and psychosis are more prevalent than their clinical counterparts and cluster together. The risk factors for psychosis may facilitate the formation of more "toxic" combinations of subclinical mania and subclinical psychosis with a higher probability of transition to bipolar disorder. A better understanding of this pathway is crucial for the development of early interventions.