This study was conducted to investigate the frequency and origin of discrepant assay results between two haemolytic assays which both measure activity of the classical pathway of complement (CH50) by haemolysis of sheep red blood cells (SRBCs). One is conducted in gel phase using undiluted sera and the other in liquid phase with sera in 1/100 dilution. The majority of discrepant readings are observed as low or absent haemolysis in the gel phase, with values within or above the normal range in the liquid phase. The incidence of discrepant assay readings was evaluated in 300 samples. Furthermore, 28 samples showing the most discrepant readings were investigated further for disturbing factors. Factors evaluated in the test sera were mannose-binding lectin, C-reactive protein (CRP), immune complexes, antibodies to SRBCs, rheumatoid factor and immunoglobulin A (IgA) and IgG anti-C1q antibodies. The results showed that discrepant readings are present in 10% of the 300 samples and false low gel assay readings account for 6.3%. The majority (68%) of the discrepant samples contained a heat-stable-inhibiting factor, and the main mediators found were elevated levels of IgA anti-C1q antibodies and antibodies to SRBCs. This could indicate a clinically relevant factor in the test sera but can also result from the difference in assay design.