Objectives: The objectives of this study are to investigate how many advanced cancer patients became unconscious or non-communicative after pharmacological treatment for delirium, and to explore whether existing delirium assessment tools can successfully evaluate its severity at the end of life. Methods: This was a secondary analysis of a registry study that examined the efficacy and safety of antipsychotics for advanced cancer patients with delirium. A total of 818 patients were recruited from 39 specialized palliative care services in Japan. The severity of delirium was measured using the Richmond Agitation-Sedation Scale-Palliative care version, the Delirium Rating Scale-Revised-98 (DRS-R-98), and the Nursing Delirium Screening Scale (Nu-DESC) on Day 3. Data from 302 patients with motor anxiety with an Agitation Distress Scale score ≥2 on Day 0 were analyzed for this study. The patients were categorized into four treatment response groups: complete response (CR: no agitation and fully communicative), partial response (PR: no/mild agitation and partially communicative), unconscious/non-communicative (UC), and no change (NC). Results: On Day 3, 29 (10%; 95% confidence intervals [CI], 7-13) and 2 (1%; 95% CI, 0-2) patients became unconscious and non-communicative, respectively. Forty-four patients were categorized as CR, 97 as PR, 31 as UC, and 96 as NC. The scores of the DRS-R-98 and Nu-DESC in the UC group were rated higher than patients in the NC group were. Conclusions: A considerable number of cancer patients with delirium became unconscious or non-communicative. Existing delirium assessment tools may be inappropriate for measuring the severity of delirium in end-of-life.