This study had two purposes: to test the usefulness of partial order scalogram analysis with multivariate response data; and to illustrate the multidimensional nature of suicide risk. A detailed introduction describes partial order scalograms, which locate respondents' profiles in a two-dimensional space (rather than on a unidimensional Guttman scale), portraying both quantitative and qualitative variation among profiles. Three groups responded to the Suicide Probability Scale (Cull & Gill, 1982): (1) a normative group from San Antonio, Texas; (2) psychiatric inpatients; (3) persons who had made a potentially lethal suicide attempt in the last 48 hours. Of 1,158 respondents, a stratified sample of 100 profiles was analyzed by the POSAC-I (Partial Order Scalogram Analysis with Base Coordinates) scaling program, a computer program from the Guttman-Lingoes series for portraying partial order relations in a two-dimensional space. A subset of items (7, 24, 25, 30, 32) representing intrapersonal behaviors, derived from the primacy-of-environment facet, yielded a solution with .93 goodness-of-fit. Spatial regions discriminated suicidal and nonsuicidal groups, but did not separate normals from psychiatric patients. There was no typical suicide profile; scalograms showed qualitative heterogeneity of suicide attempters, not evident from total scores alone. Affective expressions of hopelessness and despair and cognitive planning of means were orthogonal dimensions of suicide risk. Numerous figures aid in conveying method and results. (LPG)