Precise and accurate measurements of Nd concentrations and isotopic compositions of small sample masses (<= 30 ng Nd) are increasingly necessary in the geosciences. Here, we present a combined chemical separation and TIMS (thermal ionisation mass spectrometry) NdO+ method adapted for fossil deep-sea corals, which represents a significant improvement in precision compared to existing methods, with additional assessment of potential contamination of coral aragonite by external encrustations. After digesting and spiking samples, we purified sample Nd using RE and Ln-spec separation chemistries, and tested the efficiency of these on aliquots of USGS BCR-2. In addition, we prepared a matrix-matched in-house coral reference material to monitor the analytical precision on coral carbonate. Using a micro-loading technique and TaF5 activator on single W filaments, ion beam sizes of up to 0.60 V per ng on mass 160 are achievable, but decline with increasing age of the activator. The long-term average for NdO+ measurements of 5 ng and 15 ng loads of the pure metal oxide JNdi-1 reference material (Geological Survey of Japan) was 0.512106 +/- 6 for Nd-143/Nd-144 (12 ppm 2RSD, n= 44), with an overall deviation from the reference value of 17 ppm (0.512115 +/- 7; Tanaka et al., 2000). BCR-2 aliquots of 10 ng and 30 ng generated an average Nd-143/Nd-144 of 0.512643 +/- 8 (16 ppm 2RSD, n = 14), which is identical within uncertainty to the reference value (0.512637 +/- 12; Weis et al., 2006). Very similar external reproducibility was obtained on 10 ng and 30 ng aliquots of the in-house coral standard, with an average 143Nd/144Nd of 0.512338 +/- 8 (16 ppm 2RSD, n = 13). Precision and accuracy of Nd oxide measurements depend on well-characterised oxygen isotopic compositions over the range of measurement temperatures, and very small amounts of Ce and Pr in final chemistry cuts. Whilst the former is laboratory-specific, and can be determined precisely using a column-purified Nd-150 spike (e. g. O-17/O-16 = 0.000390 +/- 3 and O-18/O-16 = 0.002073 +/- 8 for this study; n= 12), our method shows that interference corrections for Ce and Pr are tolerable up to higher values than previously reported (e. g. 156/160 ratios <= 0.148 and 157/160 ratios <= 0.68). In reality, however, our separation was always better than this. An important issue with fossil corals concerns their apparently higher Nd concentrations compared to modern counterparts (similar to 7-310 ppb vs. similar to 3-51 ppb respectively). Variable to high Nd concentrations could arise for a number of reasons, but here we focus solely on evaluating the effectiveness of coral cleaning to remove external, Nd-rich encrustations. To test this, we measured major and trace metal concentrations (Ca, Fe, Mn, Ti, Al, Ce, Nd, Th, U) in paired aragonite and FeMn coatings from the same samples. Mass balance calculations reveal negligible influence of FeMn coatings on the final epsilon(Nd), although four samples showed some degree of contamination as demonstrated by elevated Fe, Mn, Ti and Th concentrations.