37 outpatients with at least one prospectively observed manic or hypomanic episode comprised a sample for comparison of five definitions of dysphoric (hypo)mania. Dysphoric symptoms were continuously rather than bimodally distributed. Prevalence of dysphoria varied from 5 to 73% depending on the definition used. Female gender was associated with dysphoria under two of the five definitions. Inter-episode stability in patients with at least two prospectively observed episodes (n = 15) was not significantly different from chance. These data do not indicate that (hypo)mania can be dichotomized on the basis of dysphoria. Advantages and disadvantages of dimensional and categorical approaches to specifying mood in mania or hypomania are discussed.