A protocol is described for the accurate and precise determination of Sr isotope amount ratios in the TORT-3 candidate lobster hepatopancreas certified reference material (CRM) from the National Research Council Canada. Parameters affecting the separation of Sr and Rb based on a double pass of the dissolved sample matrix through a column packed with Eichrom Sr Spec resin were investigated in detail. Matrix elements were removed and the Rb concentration was reduced to below 0.002% (expressed as the 85Rb/88Sr intensity ratio) in the purified solutions of TORT-3. Fractionation on the column was eliminated as quantitative recovery (98 ± 4% mean and 1 SD, n = 18) of strontium was obtained. A modified method of standard-sample bracketing with internal normalization for mass bias correction was employed using natural Zr added to both the sample and a solution of NIST SRM 987 used as the bracketing standard. Each iSr/86Sr ratio (88Sr/86Sr, 87Sr/86Sr and 84Sr/86Sr) was used to separately calibrate an individual 91Zr/90Zr ratio, which was then used to correct corresponding 88Sr/86Sr, 87Sr/86Sr and 84Sr/86Sr ratios in the sample. The interference from residual 87Rb on 87Sr was corrected for by simultaneous measurement of 85Rb and the use of the IUPAC 87Rb/85Rb ratio of 0.385706, which was itself mass bias corrected assuming the same mass bias for 87Sr/86Sr and 87Rb/85Rb. SRM 987 standard solutions used for method validation returned ratios in good agreement with their certified values, confirming the accuracy of the approach. Absolute isotope amount ratios of 0.056526 ± 0.000048, 0.70937 ± 0.00010 and 8.3824 ± 0.0012 (U, k = 2) for 84Sr/86Sr, 87Sr/86Sr, 88Sr/86Sr, respectively, were obtained in TORT-3. The main contributors to the measurement uncertainties of the TORT-3 Sr isotope ratios are the certified values in SRM 987. Values of −0.442 ± 0.071‰, −1.363 ± 0.036‰ and 0.454 ± 0.030‰ (mean and 1 SD, n = 18) characterize δ84/86Sr, δ87/86Sr and δ88/86Sr in TORT-3, respectively.