Routine techniques for Sr isotope analysis on a double-focusing multiple collector inductively coupled plasma mass spectrometer (MC-ICPMS) are described that for unspiked samples yield accurate and precise results (e.g., Sr-87/Sr-86 = 0.710255 +/- 24; 0.003%; 2 S.D.) similar to those obtainable by thermal ionisation mass spectrometry. Prior to a 5-min analysis, backgrounds are measured on-mass to correct for interferences on masses 84 and 86 derived from small amounts of Kr in the Ar supply, and stable, residual memory Sr and Rb signals from sample material on the torch and cones. Each analysis consumes ca. 300 ng of Sr, which compares unfavourably to TIMS where loads of ca. 1 ng can be measured. Simultaneous ionisation of all elements during plasma-based ionisation proves to be a disadvantage during Sr isotope analysis as extra chemical clean-up steps, beyond those necessary for TIMS analyses, are required to remove potential isobaric interference problems such as Rb, doubly charged rare earth element (REE) and Ca dimers and argides. Reproducibility for spiked Sr samples is significantly worse than for unspiked samples, largely due to an inability to determine the exact isotopic composition of interfering Rb. This problem can be avoided by splitting the digested sample into isotopic composition and dilution aliquots prior to spiking. Laser ablation (LA)-MC-ICPMS Sr isotope ratio measurements of various geological and biological samples have shown significant deviations in the naturally constant Sr-84/Sr-86 ratio and the radiogenic Sr-87/Sr-86 ratio from values obtained by analysis of Sr chemically separated from the sample matrix. The precise reasons for this remain unclear but likely reflect a combination of isobaric interferences from Ca dimers and Ca argides and doubly charged REE, as well as disruption of mass bias effects due to differential loading of the plasma. It is also unclear whether this is an instrument- or technique-specific problem and we suggest great care be taken when attempting Sr isotope analyses by in situ LA-MC-ICPMS techniques. (C) 2002 Elsevier Science B.V. All rights reserved.