Background Following the introduction of two-site immunometric assays for parathyroid hormone (PTH), the expectation of good inter-assay agreement has not been fulfilled. The reasons for this may include differences in standardization as well as fragment recognition between the assays.Methods PTH values for healthy individuals, patients with renal failure and patients with normal renal function and elevated parathyroid hormone (hPTH) were compared using two commercial two-site immunochemiluminometric assays (Bayer Magic-lite(R) and DPC Immulite 2000(R)).Results Immulite results had a mean value 50.4% greater than the corresponding Magic-lite values for the whole study population with individual values ranging from 17.5% below to 118.3% above the corresponding Magic-lite value. There was no significant difference in inter-assay bias between patients with renal failure and those with normal renal function, suggesting that variable cross-reactivity with circulating disease-specific PTH fragments was not the primary cause of the observed discrepancy. Cross-reactivity with the synthetic fragment hPTH (7-84) was 34 +/- 5% for Magic-lite and 62 +/- 2% for Immulite. We also studied the stability of synthetic hPTH on storage.Conclusion The instability of synthetic hPTH over extended storage periods may affect primary standard material. The consistent inter-assay differences and the over-recovery observed in external quality assessment programmes for the Immulite assay may have best been explained by differences in calibration and the relative cross-reactivities and/or kinetics of the two assay systems for specific parathyroid fragments.