The enzymatic degradation of cocaine in blood samples, even during transport to a forensic laboratory, is a common problem in toxicological analysis. This can be avoided by the use of blood-sampling devices such as gray-top Vacutainers containing the cholinesterase inhibitor sodium fluoride. In the present study, which included 147 authentic cases, blood samples were collected into two different tubes, one containing fluoride/oxalate and one without stabilizing agents. In all cases, both samples were analyzed for drugs of abuse using Abbott FPIA immunoassays after precipitation and gas chromatography-mass spectrometry (GC-MS) for quantitative analysis. The cannabinoid immunoassay showed markedly lower values in the fluoride-containing samples; this was investigated further and could be explained by hemolysis of these samples. In addition, the concentrations of 11-nor-delta9-tetrahydrocannabinol-9-carboxylic acid (THCCOOH) were lower in these samples. A stability study with the THCCOOH acyl glucuronide showed that it is unstable in unpreserved serum, which could explain our observation. GC-MS quantitative data for amphetamine and derivatives, opiates, delta9-tetrahydrocannabinol, and 11-hydroxy-delta9-tetrahydrocannabinol were essentially identical; however, they also differed substantially for cocaine, cocaethylene, ecgonine methylester, and benzoylecgonine. Unexpectedly, the concentrations of benzoylecgonine in unpreserved serum were almost half as high as in the fluoride-containing samples.