We recently published in this journal an article entitled ‘‘Plasma oxidative stress biomarkers, nitric oxide and heat shock protein 70 in trained elite soccer players’’ (Banfi et al. 2005) but our results with the d-ROMs test (Diacron International, Grosseto, Italy) were believed invalid by Harma et al. (2006), who reported the findings of a previous paper of Erel (2005) comparing the above test with the new iron-o-dianisidine/xylenol orange assay. In this article we provide the following evidence strongly indicating that the above authors made some relevant errors in the assessment of the d-ROMs test validity. Firstly, the relationships between d-ROMs test and ferroxidase activity are neither new nor an original finding. Indeed, Alberti et al. (2000), just in the study leading to the definitive validation of the d-ROMs test by electron paramagnetic resonance (EPR) spectrometry, reported that the contribution of ceruloplasmin ferroxidase activity to the typical change of absorbance at 505 nm (DA505/min) in such a test (see below), although not negligible, appeared relatively small. This is because in the d-ROMs test the serum sample is 100-fold diluted. Moreover, Alberti et al. (2000) demonstrated that an amount of sodium azide equivalent to the maximum expected concentration of ceruloplasmin in the serum (i.e. 4.0 lM) causes only an approximately 7% decrease of the DA505/min value. This may explain the reported correlation between d-ROMs test results and ferroxidase activity (Erel 2005). Of course, the above correlation does not mean that d-ROMs test measures only and exactly the serum ferroxidase activity. Indeed, a 70% of the DA505/min value is still detectable even with a five-to-ten fold excess of the azide in the test (Alberti et al. 2000). Furthermore, the found correlation between ferroxidase activity and d-ROMs test (Erel 2005) is not a general rule. For instance, in hemodyalised patients d-ROMs test value was shown to be increased (Gerardi et al. 2002) while the ferroxidase activity of ceruloplasmin was reduced (Roxborough et al. 2000). In this respect, if right that d-ROMs test measures only the ferroxidase activity, Erel (2005) failed to demonstrate and/or explain how an increased ceruloplasmin activity may justify the experimental/clinical findings, which have been reported to be associated to an increased result of d-ROMs test. On the other hand, even granting, for the sake of an argument, that the above relation is the rule, this is not consistent with the recently recognised controversial role of ferroxidase (Shukla et al. 2006). About the lack of response of d-ROMs test during copper-induced lipoprotein autoxidation, apart from the questionable significance of this experimental approach, this finding may be explained by the fact that d-ROMs test is significantly inhibited by the adding of chelants (Alberti et al. 2000) which sequester the iron thus, inhibiting the Fenton’s reaction. Indeed, Erel (2005), in the autoxidation tests used a citrate buffer to obtain lowdensity lipoprotein and ethylendiamine tetra acetate (EDTA) to prevent further oxidation during the eating. This reply refers to the Letter to the Editors, at http://dx.doi.org/ 10.007/s00421-006-0202.