Several small studies suggest a link between environmental exposure to organochlorine compounds and risk of non Hodgkin's lymphoma (NHL). Because NHL is uncommon, studies of the topic often use a population-based case-control design, in which cases generally are enrolled after treatment has begun. If chemotherapy affects blood levels of organochlorines, exposure will be misclassified and findings distorted. To determine whether chemotherapy alters serum levels of organochlorines in NHL cases, we compared serum samples before and after treatment in 22 cases diagnosed with NHL between March 1994 and August 1995 and enrolled in a clinical trial at the United States National Cancer Institute's Clinical Center. The time difference between pretreatment and posttreatment samples ranged from 15 to 27 months with an average of 20 months. Laboratory analyses were conducted in blinded pretreatment and posttreatment pairs of the subjects. Pretreatment and posttreatment organochlorine serum levels were compared using Pearson correlation coefficient (r) and paired t test. The pretreatment and posttreatment serum levels were highly correlated for 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene (DDE) and polychlorinated biphenyls (PCBs) PCB-138, PCB-153, PCB-156, and total PCBs (ranging from 0.78 to 0.93). Serum levels of all of these organochlorines significantly decreased between initiation and completion of chemotherapy, 25% for total PCB (P = 0.0044), 28% for DDE (P = 0.0014), 25% for PCB-138 (P = 0.0053), 27% for PCB-153 (P = 0.0031), and 29% for PCB-156 (P = 0.045). Neither weight change nor lipid change was correlated with changes in chemical levels. There was no association between the length of time between blood draws and changes in chemical levels. Our data raise the possibility that lymphoma treatment depresses serum organochlorine levels.