In a collaborative study involving three laboratories, randomly coded sera from 47 patients and healthy donors were tested for soluble immune complexes by two versions of the C1q BT and by the C1q DT. Analysis of ranked data showed a close correlation between results obtained in all laboratories as well as good reproducibility on testing coded duplicate samples included in each panel of sera. However, with the use of values for normal donors established independently in each laboratory, the tests did not always agree in discriminating normal from abnormal sera, particularly with sera from cancer patients. Results reflected to some extent the methods used to inactivate the endogenous C1 complex in the test sera. Studies of 130 additional cancer patients revealed that 44 (34%) gave abnormal C1q BT values when inactivated with EDTA but only 35 (27%) were abnormal when heat inactivated (56 degrees for 30 min). Similarly 12 of 65 sera (18%) from patients with nonneoplastic diseases and 10 of 80 (13%) from healthy donors gave significantly abnormal results after addition of EDTA. Eight (12%) of the disease and six (8%) of the healthy controls' sera were similarly outside normal limits when heat inactivated. Repeated freezing and thawing of sera or changes in the concentration of PEG used to precipitate complex-bound 125I-C1q influenced C1q BT results more than duration of storage at -70 degrees C or changes ascribed to presence or absence of rheumatoid factors and CRP.