OBJECTIVE: We studied 405 patients evaluated by the Psychosomatics Service to investigate whether the behavioral description as reflected on consultation request from primary medical-surgical teams could be utilized to predict the final clinical diagnosis of delirium. We explored whether outcomes differed in patient with delirium with different consultation requests. METHODS: Patients with delirium (n=114) were divided into subtypes based on consultation requests for the management for cognitive dysfunction [altered mental status (AMS), n=46], for agitation or behavior disturbance (BEH, n=26), for depression (DEP, n=18), and for all other reasons (MISC, n=33). Adjusted multivariate regression models were used to examine group differences. RESULTS: Consult requests for AMS had a high likelihood ratio (LR+=14.22) and requests for BEH had moderately high likelihood ratio (LR+=4.79) of receiving a diagnosis of delirium. The DEP group (n=18) tended to be younger and had higher 30-day readmission rates. The BEH group (n=26) had more comorbid systemic medical conditions, higher rate of being discharged to home, and lower mortality rate. Delirium subtypes did not show any significant difference in clinical outcomes. CONCLUSION: Behavioral observations on consultation requests as formulated by primary medical-surgical teams may be useful in the classification of patients with delirium. Whether the behavioral observations on the request may be used to predict delirium outcomes deserves further research.