The reconstruction of past variations in seawater strontium isotope ratios (Sr-87/Sr-86) from marine carbonate skeletons (MCSs) is based on the assumption that the Sr isotope composition of oceans is homogeneous worldwide. However, fossil materials are mainly from shelf domains whose modern Sr budget is known to be very sensitive to continental Sr fluxes. Their reliability as chemostratigraphic tools or weathering proxies may thus be questioned, especially because no formal appraisal of biological and environmental biases has been done on a global scale. Here, we provide new modern MCS Sr-87/Sr-86 values from various marine environments that we analyze statistically in regards to extensive compilations of MCS and water data. We show that most MCS Sr-87/Sr-86 values fall within the range of those of modern oceanic waters (0.709172 +/- 23). Others, concerning similar to 30% of studied localities, may display slight to considerable offsets compared to the Sr-87/Sr-86 signature of modern seawater (+/- 250 x 10(-6)). These patterns mainly concern bivalves and gastropods, or more generally epibenthic and eurytopic organisms, from coastal domains with water mass restriction, low salinity, or strong continental supplies (fjords, lagoons, endorheic basins). Beyond classical fluvial freshwater influences, submarine groundwater discharges may lead to a less radiogenic Sr-87/Sr-86 signature of MCSs even in lagoons with euhaline conditions (e.g., Oualidia lagoon, Morocco). All biases considered, this calls for caution in relative dating, as we calculate that half the values of the Phanerozoic Sr-87/Sr-86 curve cannot provide time calibration accuracy better than +/- 3 m.y., which is necessary to discriminate the shorter stages.