Abstract Altered concentration of the vertebrate yolk protein precursor vitellogenin is a recognized biomarker for endocrine disruption in fish, and within recent years yolk protein alteration has also been associated with endocrine disruption in bivalves. Species-specific, direct and sensitive methods for quantification of vitellogenin in fish have been available for years whereas bivalve yolk protein levels have been estimated indirectly by alkali-labile phosphate (ALP) liberated from high molecular weight proteins because the sequence and biochemical structure of most bivalve yolk proteins are unknown. By applying a species-specific enzyme-linked immunosorbent assay (ELISA) for accurate determination of yolk protein level the impact of 17β-estradiol (57, 164 and 512ng/L) on the freshwater bivalve Unio tumidus was investigated and compared with ALP estimations. Seven weeks of exposure during the pre-spawning and spawning period had no consistent effect on yolk protein concentration in hemolymph, and ALP levels in hemolymph also remained unchanged in both males and females. Further, basal male and female ALP levels were indistinguishable whereas the ELISA demonstrated that yolk protein levels of females exceeded male levels at the time of sampling, although male basal levels were high compared to fish. Altogether it is shown that individual ALP levels do not reflect yolk protein levels and hence hemolymph ALP levels cannot serve as biomarker for estrogenic exposure during the pre-spawning and spawning period in U. tumidus. The necessity of sensitive and validated biomarkers for reliable interpretation of data and the utility of ALP and yolk protein levels as biomarkers in bivalves are discussed.