Dengue, although endemic in Puerto Rico, is often not mentioned in the death certificates of decedents with laboratory results positive for dengue. Because confirmatory results are usually not available during hospitalization, we examined the utility of 2 instruments for diagnosis on the basis of clinical findings: the definition of dengue hemorrhagic fever (DHF) and the publicized (but unevaluated) clinical alarm signals for impending dengue shock. We studied data from all patients with laboratory test results positive for dengue who died (23 patients) and from the 8 patients whose death certificates listed dengue as a cause of death but whose laboratory test results were negative for dengue in Puerto Rico from 1992 through 1996. We examined hospital records to determine whether the clinical criteria for DHF were fulfilled and evaluated the incidence and timing of clinical alarm signals (intense, sustained abdominal pain; persistent vomiting; sudden change from fever to hypothermia; and marked restlessness or lethargy) and the hematocrit/hemoglobin ratio as an indicator of hemoconcentration. A similar proportion of patients with laboratory test results positive for dengue (18 [78%] of 23) and negative for dengue (6 [75%] of 8) fulfilled the criteria for DHF. Clinical alarm signals were found only among patients with laboratory test results positive for dengue and were usually noted on the day that the patient's condition deteriorated. The hematocrit/hemoglobin ratio identified 1 (6%) of 16 patients with dengue who had significant hemoconcentration. Important comorbidities were present in 16 (70%) of the patients with laboratory test results positive for dengue and in 4 (50%) of the patients with dengue-related deaths with laboratory test results negative for dengue. Dengue-related deaths in Puerto Rico often occur in patients with comorbidities. Among such patients, the DHF definition and the hematocrit/hemoglobin ratio were not useful in identifying patients with laboratory test results positive for dengue. In contrast, the clinical alarm signals for shock supported the dengue diagnosis and should alert clinicians to the severity of the disease.