A prothrombin-time-derived method was used to measure plasma fibrinogen concentration (PFC) in 286 samples from 242 normal and 44 orally anticoagulated subjects. Absorbance changes at 405 nm (ΔOD) during the clotting process were obtained by an automatic coagulometer and their relationship with plasma fibrinogen concentration (range 90–1090 mg/dl), measured by the Clauss method, was investigated. A weighted linear regression between the ΔOD and the Clauss-derived PFC values provided the best fit of the experimental data. The fitting equation was found to be reliable and accurate for PFC determination in normal subjects, whereas a systematic overestimate of fibrinogen level was demonstrated in plasma with high fibrinogen concentrations (> 400 mg/dl) and in plasma from anticoagulated patients. The systematic overestimate in the latter samples could be a result of an increased fibrin gel turbidity, as shown by in-vitro experiments using purified fibrinogen clotted by different thrombin concentrations. The PFC overestimate by the prothrombin-time-derived method could also be experimentally reproduced by competitively inhibiting thrombin–fibrinogen interaction by hirudin 54–65 peptide and the fibrinogen fragment E. A similar qualitative result was also found for the prothrombin-time-derived method in the presence of the Gly-Pro-Arg-Pro peptide, which competitively inhibits the end-to-end fibrin aggregation process. Notably, under both the above experimental conditions, the Clauss method underestimated the PFC. On the other hand, the 'clot recovery' method was minimally affected by the above inhibitors. These results indicate that the prothrombin-time-derived method is accurate and precise for most routine purposes. Its precision seems inadequate, however, under those conditions where the prothrombin time is prolonged (such as anticoagulant therapy) and in the presence of high fibrinogen levels.