The influence of C3 levels on the determination of C3d in plasma and synovial fluid by radial immunodiffusion was investigated. In the method used, C3 is precipitated by 11% polyethylene glycol (PEG), and C3d is measured in the supernatant. In 51 healthy donors, a weak though significant correlation between C3 and C3d levels was found. The mean concentration of C3d was 1.6% of that in aged serum from healthy donors. So, small amounts of C3 (i.e., 1-2% of the normal plasma level) in the 11% PEG supernatants may contribute significantly to the C3d levels measured. A radioimmunoassay that detects C3, C3b, iC3b and C3c was used to measure C3 levels in the PEG supernatants. In PEG supernatants of 4 plasma samples, 0.3-0.6% of the C3 level in normal plasma was found, whereas in those of 2 synovial fluids much higher levels were found (4-10% of the normal plasma level). When purified 125I-labeled antibodies against C3c were added to the gel of the radial immunodiffusion, C3c antigen was detected in the precipitation rings obtained with all PEG supernatants of plasma samples from patients. Therefore, the quantitative contribution of C3 to the precipitation rings in the C3d radial immunodiffusion was analyzed after the addition of an excess of anti-C3c antibodies to the gel. No effect on the size of the C3d-precipitation rings obtained with plasma samples from patients was observed. However, the C3d precipitation rings obtained with synovial fluids were significantly smaller when the gel used in the radial immunodiffusion contained an excess of anti-C3c antibodies together with the anti-C3d serum. We conclude that it is necessary to add an excess of anti-C3c antibodies to the gel used for the radial immunodiffusion, for the determination of C3d levels in synovial fluid. An antiserum against human C3b, which contains both anti-C3c and anti-C3d antibodies, can be used for this purpose.