The assessment of an individual's risk f m suicule is one of the clinician's most challenging tasks. Recent advances in both research and theory on suicide offer s m guidelines to those who work with suicidal patients, including viewing risk assessment us an ongoing feature of treatment. Assessment i s presented in a two-tiered model comprising backgroundlcmtextual factors and subjectivity. The clinical assessment of subjectivity is fmulated around Shneidmn's (1985) concepts of perturbation and lethulity. The decision regarding hospital admission versus umbulutory cure i s also discussed. A theoretically infmmd approach to assessment should serve both interview and m e traditional psychological assessment methods.Prevention rests on assessment; assessment rests on definition.Shneidman (1985, p. vi) Of the common categories of suicidal behaviors, completed suicide is the most difficult to predict; it is the least common. Whereas the lifetime prevalence of attempted suicide in the general population in North America is estimated at 3% (Moscicki et al., 1989;Sakinofsky & Webster, 1994), the lifetime prevalence of completed suicide, at a conservative annual mean rate of 10 per 100,000. is 0.5%. A related and classic problem is the extrapolation of characteristics that denote suicidalit); in high-risk populations to specific individuals in those populations. Maris, Berman, and hlaltsberger ( 1992) have identified several other salient issues in suicide risk assessment, including membership in a high-risk group, the acuteness of risk/probability of action, the need for risk indicators to be clinically relevant, and the fact that "what is being predicted or assessed is not one thing, but actually many niul t id i men s ional , intersecting , and interacting para meters" ( p p.642-633). I n searching for relevant risk factors, the clinician should be cautious about combining clinical features found arnong suicide ideators, attempters, and completers, because they appear t o be o\.erlapping but different populations. The question of suicide risk is thus becoming more refined: Risk for which suicidal behavior, exactlv? 3lodels of risk assessment have often taken the form of lists of discrete symptoms or patterns, without an integrated perspective or guidelines for implementation. An assessment model should help organize the clinician's thinking about specific questions regarding a particular patient. It should be based, however, on current understanding of these questions. We propose a twotiered model of suicide risk assessment that comprises both background and subjective risk factors. 'The background factors are the sociodemographic and related indices that have been found to be correlated with increased risk. These indices are based on different cohorts, populations, and cultures, and many of them change over time (we describe North American trends). Backgrouncl factors are once removed from the individual being assessed. but in the aggregate they can inform the clinician about a patient's general level of risk. T...