Neodymium (Nd) stable isotopes have the potential to provide new constraints on a diverse range of geological processes from planetary formation and magmatic differentiation to weathering and ocean circulation. In this contribution, we present a technique for the high-precision measurement of Nd stable isotope ratios by thermal ionisation mass spectrometry (TIMS). We use a 145Nd-150Nd double spike (DS), composed of 28% 145Nd and 67% 150Nd, to correct for mass dependent fractionation resulting from sample preparation and mass spectrometry. Isotope ratios are expressed as δ146Nd which is the per mil deviation in the measured 146Nd/144Nd ratio relative to reference material JNdi-1. Repeated analyses show that an internal precision (n = 400 cycles; 2 se) of ≤0.005‰ on δ146Nd can be achieved in agreement with theoretical predictions. An interlaboratory comparison of the primary standard JNdi-1 over a three-year period shows that resolvable offsets in δ146Nd are induced within and between mass spectrometers as the result of systematic biases in the efficiency of the Faraday collectors. Following normalisation of the data to JNdi-1 = 0‰ the long-term reproducibility of δ146Nd on a range of international reference materials processed through chemical separation is better than ±0.015‰ (2 sd). In addition to stable isotope compositions this method allows for the simultaneous measurement of 143Nd/144Nd with a precision of ≤11 ppm. Over a wide range of 143Nd/144Nd (0.5118-0.5132) the DS analyses here agree within analytical uncertainty with published values using conventional techniques. To allow the accurate age correction of these radiogenic isotope compositions in ancient geological samples (>2 Ga) here for the first time we have combined a DS and an elemental tracer spike. A 149Sm tracer spike was calibrated and used to obtain isotope dilution Sm concentrations and ultimately parent-daughter ratios. Measured Sm-Nd ratios are within uncertainty of previous estimates for a range of reference materials, with the addition of the Sm spike having no resolvable effect on δ146Nd values. This technique allows δ146Nd, 143Nd/144Nd and Sm/Nd to be obtained simultaneously, therefore allowing constrains to be placed on both the source or age of a material and the processes involved in its formation.