BACKGROUND: One of the most serious risks of epidural anesthesia is total spinal blockade from unintentional dural puncture. We evaluated the glucose test and the thiopental precipitation test to differentiate cerebrospinal fluid (CSF) from local anesthetics (LA). METHODS: (1) Experiment 1: CSF from twenty patients was serially diluted with 2% lidocaine or 0.5% bupivacaine. The ratio of CSF to LA-CSF mixture (CSF/(LA+CSF)) was from 0 to 1.0 at an interval of 0.1. We measured the glucose level of each sample with blood sugar meter. (2) Experiment 2: CSF from a hydrocephalus patient was serially diluted and its glucose level of each sample was measured in the same way as Experiment 1. We performed a urine stick test with each sample. Ten anesthetists blinded to the nature of the sample were asked to identify the results of the tests. (3) Experiment 3: Two milimeters of 2.5% thiopental was respectively mixed with local anesthetics, the amount of which was from 0.1 to 1.0 ml at an interval of 0.1 ml. Sixteen anesthetists blinded to the nature of sample were asked to identify the results of the tests. RESULTS: (1) Experiment 1: We can measure glucose level at CSF/(LA +CSF) of 0.5 in 2% lidocaine group and 0.6 in 0.5% bupivacaine group. (2) Experiment 2: We can detect glucose at lower level of CSF/(LA +CSF) by glucose meter than urine stick test (p<0.05). (3) At least 0.35 ml of 2% lidocaine and 0.29 ml of 0.5% bupivacaine was needed respectively to detect precipitation. CONCLUSION: We suggest that blood glucose meter be used instead of glucose test strip. For thiopental precipitation test, we have to adjust the amount of thiopental depending on the amount of test fluid.