Our aim was to explore the agreement between clinically collected information on purported drug intake and plasma data in intentional drug overdose. We included all subjects with intentional drug overdose above 15 years of age consecutively admitted to the Emergency Department of the University Hospital during 4 months. Information about drugs used and sources of this information was collected and compared to presence of drug in plasma, concerning four drugs with high toxic potential (tricyclic antidepressants, meprobamate, paracetamol and ethanol). Sensitivity, specificity, predictive positive and negative values of all sources of information pooled were assessed for each drug. 413 intentional drug overdoses were included, 66% with more than one drug. According to clinical information, 8% took tricyclic antidepressants, 11% meprobamate, 9% paracetamol and 41% ethanol. Systematic plasma assays confirmed this in 59% of cases for tricyclic antidepressants, 76% for meprobamate and ethanol, and 77% for paracetamol. Plasma concentrations were considered toxic in 28% of cases for tricyclic antidepressants, 65% for meprobamate, 43% for ethanol and never for paracetamol. Tricyclic antidepressants and meprobamate were found unexpectedly in 3%, paracetamol in 7% and ethanol in 6%. Toxic concentrations were found only with meprobamate. The risk of erroneous, clinically collected information was greater by excess (25 to 40% false positives) than by lack (3 to 7% false negatives). Thus, the consequences of erroneous, clinically collected information were probably more excess cost for the institution than medical risk for the patients. However these results found at the population level may not be true at an individual level.