We were unable to validate the predictive ability of a previously described mneumonic (SAD PERSONS score, SPS) for determining the need for hospitalization of patients who have expressed suicidal ideation or behavior. After reviewing published studies on the subject, we arrived at a modification of this scale (MSPS). The MSPS obtained by nonpsychiatric and psychiatric house staff was compared and correlated with patient disposition for 100 consecutive adult patients presenting to the Harbor-UCLA Medical Center emergency department for the assessment of suicidal behavior. We were unable to correlate the MSPS with patient disposition. However, four criteria were found to correlate with the need for hospitalization (P less than .001). After weighting the MSPS, giving two points for each of these four criteria, and one point for each of the other six, we found that a score of greater than or equal to 6 had a sensitivity of 94% and a specificity of 71% in identifying need for hospitalization. A score of less than or equal to 5 had a negative predictive value for hospitalization of 95%. When two nonsuicidal patients, admitted for situational reasons only, were eliminated, a score of greater than or equal to 6 had a 100% sensitivity and a score of less than or equal to 5 had a 100% negative predictive value in identifying patients requiring hospitalization. Subsequent use of the weighted MSPS for an additional 82 patients confirmed these results. Follow-up of 97.5% (116/119) of all patients discharged from the emergency department at 6 to 12 months found no mortalities. Using a weighted MSPS, nonpsychiatrists can quickly and easily obtain the objective information necessary to make an initial assessment of suicidality.(ABSTRACT TRUNCATED AT 250 WORDS)