Limestone–marl alternations are usually directly interpreted to reflect cyclic palaeoenvironmental signals. However, uncertainty in such interpretations stems from the differential diagenesis that most limestone–marl alternations have undergone. Differential diagenesis results in markedly different alterations between limestones and marls and in the loss of comparability of many measurable parameters. For an unequivocal interpretation of the origin of rhythmic alternations, diagenetically robust parameters or parameters that clearly indicate the degree of diagenetic bias are required. The present study uses a multiproxy approach (independent biotic, sedimentary and geochemical parameters) in order to unravel the palaeoenvironmental signal recorded in Valanginian (Early Cretaceous) limestone–marl alternations from the Blake-Bahama Basin (DSDP site 391). Using this approach, terrestrial and marine influences can be distinguished, changes in nutrient levels estimated and prediagenetic differences in the non-carbonate fraction constrained. Surprisingly, no systematic variations in any of these parameters were observed between limestone and marl layers, implying that none of these was directly responsible for the formation of the rhythmic alternation. Hence, none of the current models of sedimentary formation of limestone–marl rhythmites is applicable here. Calcareous nannofossils are equally well preserved in limestone and marl layers, ruling out their dissolution in marl layers as a source of the calcite cement in the limestone beds. Sr values of 700–900 p.p.m. indicate that aragonite may have been present in the original, pelagic sediment. The assumption of fine-grained sedimentary aragonite imported from nearby carbonate platforms as the source of the cement would explain a number of otherwise enigmatic features in these rhythmites, including the source of the calcite cement observed in the limestones, the equally good preservation of calcareous nannofossils in limestones and marls and the higher concentration of calcareous nannofossils in marl layers. The study demonstrates that examination of diagenetically inert parameters or parameters in which diagenetic effects can be filtered can yield unexpected results. Clearly, careful analysis of such parameters needs to be undertaken in order to make reliable palaeoenvironmental interpretations from rhythmite successions.