BACKGROUND Blood samples in epidemiological studies are often stored for several years and analysed at different occasions. The reagent kits are continually modified for better precision and accuracy. Our hypothesis was that epidemiological studies are affected by long-term storage and/or modifications of reagent kits. METHODS Plasma samples stored at -80( degrees )C from two populations were used: A case-referent study with samples collected from 1985 to 2000 and analysed 2005 (n=1598) were used to study influence of long-term storage. A cross-sectional study analysed 1990 (n=1558) and re-analysed 2001 (n=78) and 2005 (n=828) was used to study influence of reagent kit modifications. Fibrinolytic analyses included immunoassays of tPA, PAI-1 and tPA-PAI-1 complex and chromogenic substrate assays of the activities of tPA and PAI-1. RESULTS Long-term storage for a median time of 11.6 years (range 5 to 20) showed an effect of time on tPA antigen R(2)=0.01, PAI-1 antigen R(2)=0.01 and tPA-PAI-1 complex R(2) = 0.02. Modifications in reagent kits affected the levels of fibrinolytic factors; for tPA antigen the slope coefficients were between 0.72 and 0.95 (R(2) 0.47 - 0.75), whereas tPA activity showed an agreement with slope coefficients 1.06 to 1.09 (R(2) 0.67 - 0.93). CONCLUSIONS This study showed that long-term storage affects fibrinolytic variables to a negligible extent, but modifications in reagent kits introduced an element of bias. We conclude that analysis of samples on a single occasion is preferable to multiple occasions, as storage has negligible effect.