Stable performance between-runs are essential in longitudinal studies and when different studies are being compared. However, changes in analytical kits and laboratory material occur and have the potential to threaten analytical stability. In the present case, we examined how salivary cortisol measurements in our laboratory were affected by: 1) changes in the tampon material and 2) changes in the antibody of the analytical kit. In study 1, saliva from healthy subjects (n = 19) was split and spiked to Salivette polyester and cotton tampons, respectively, and treated as ordinary samples before being analysed for cortisol using a Spectria RIA kit for cortisol. In study 2, 68 anonymous saliva samples were analysed with the Spectria Cortisol RIA kit both before and after the manufacturer changed the antibody. The change from polyester to cotton tampons reduced the measured concentration of salivary cortisol by 62 %. A difference of 12 % between the two runs with different antibodies could not be attributed to differences in storage or in thawing and freezing of samples. To conclude, both a change in the material of the Salivette used for collecting saliva samples as well as a change of antibody in a kit can have a major impact on measurements, as illustrated here for concentrations of cortisol in saliva. It is therefore recommended always to check that the analysis stays in statistical control in one's own laboratory when changes are made, even if the manufacturer reports that the changes should have no effects.