Abstract Objective To characterize the effects of serum separation tubes (SST) on serum drug concentrations. Sample Population Clinically normal dogs (clorazepate, n = 7) or dogs with epilepsy (phenobarbital, n = 7) were studied in experiment 1, and samples submitted for therapeutic drug monitoring (n = 87) were studied in ex-periment 2. Procedure In experiment 1, blood containing either drug was placed in 2 types of 4-ml SST (SST-A and SST-B) and in nonserum separation tubes (non-SST [control]). Samples were processed, then stored at 20 to 22 C (both drugs) or 10 C (phenobarbital only). Aliquots were collected for 96 hours. The rate constant of disappearance and the percentage decrease of each drug over time were determined for each tube. For experiment 2, paired samples were collected in non-SST and SST and submitted by mail for therapeutic drug monitoring. The SST samples were either decanted from SST prior to shipment (group 1; n = 30) or mailed in SST with serum in contact with the silica gel (group 2; n = 57). Drug concentrations and drug elimination half-life were compared between groups. For both experiments, drugs were detected in samples, using polarized immunofluorescence. Results For experiment 1, the rate constant of drug disappearance for both drugs was greater in the 4-ml SST-A (P <0.0001). This SST also caused the greatest percentage decrease (20% for phenobarbital and 35% for benzodiazepines) at 96 hours. Refrigeration reduced the mean decrease in phenobarbital at 96 hours to 11%. For experiment 2, phenobarbital concentration was lower for both SST, compared with non-SST (P < 0.0005). Phenobarbital had decreased a mean 6.4 ± 0.5% in group-1 and a mean 30.5 ± 11.1% in group-2 (P < 0.0005) samples. Conclusion The SST should be avoided when collecting serum for monitoring of either phenobarbital or benzodiazepines. Clinical Relevance The SST can falsely decrease serum drug concentrations and should be avoided when collecting blood for therapeutic drug monitoring. (Am J Vet Res 1996;57:1299-1303)