OBJECTIVE The purpose of this study was to establish an early prognostic model of patients with glyphosate-surfactant (GlySH) herbicide intoxication. METHODS A case-control study was conducted. Data of GlySH-intoxicated patients were collected from 2 hospitals. Patients were admitted to the emergency departments (EDs) of Chang Gung Memorial Hospital from April 1996 to March 2003 and Taichung Veterans General Hospital from April 2000 to October 2003. Collected variables such as age, sex, estimated amount of ingestion, symptoms/signs including first vital signs, chest x-ray (CXR), and biochemical studies were analyzed for their role in the prognostic model of GlySH intoxication mortality. Univariate and odds ratio analyses were then performed. The prognostic model was then established by using logistic regression analysis and further stratified analysis. RESULTS Fifty-eight patients (19 men and 39 women; age, 48.8 +/- 15.8 years; P = .38) were enrolled in our study. Forty-one patients survived from GlySH intoxication and 17 died. After univariate analysis, 5 variables (respiratory distress needing intubation, metabolic acidosis, tachycardia, elevated creatinine (Cr) level, and hyperkalemia) were found to be highly associated with poor outcome and mortality. Then a multiple logistic regression model was established as follows: log(p/q) = -6.13 + 3.43 (abnormal CXR) + 2.53 (metabolic acidosis) + 2.55 (Cr) + 2.4 (tachycardia) + e. CONCLUSIONS GlySH poisoning is multiorgan toxicity. Pulmonary toxicity and renal toxicity seem to be responsible for its mortality. Metabolic acidosis, abnormal CXR, tachycardia, and elevated Cr level are useful prognostic factors for predicting GlySH mortality.