The consequences, possible origins, and prevention of mass-scale drifts in the high mass resolution mode (HR, M/ΔM ≈ 8000) under constant conditions were investigated and simulated in case of a multicollector-inductively coupled plasma mass spectrometer (MC-ICP-MS) using silicon enriched in 28 Si as the main element in this survey. A drifting mass scale strongly impairs the precise and accurate determination of isotope ratios, depending especially on the peak/plateau width. For example, 29 Si+ in Si highly enriched in 28 Si has an extremely small mass plateau width of ΔM ≤ 4 × 10-3 u, compare ΔM(56 Fe+ ) ≈ 18 × 10-3 u, which is to our knowledge one of the smallest plateaus routinely investigated in isotope ratio measurements, thus requiring extreme stability. During warm-up of the double-focusing sector field mass spectrometer, a mass drift up to ΔM/Δt ≥ 0.006 u/hr has been observed. Long-term studies on mass scale stability and simulations concerning fluctuations of the magnetic field B, acceleration voltage Uacc and ESA voltage UESA are reported. A change of one of these quantities of 0.01% induces changes of the mass scale of 6 × 10-3 u, 3 × 10-3 u, and 1 × 10-3 u in the case of B, Uacc , and UESA , respectively. After identifying electrical charging/discharging effects in the mass spectrometer affecting the mass scale stability, the instrument was completely dismantled and carefully reinstalled. Additional stability tests using silicon, strontium, and lead finally yielded a mass drift of ΔM/Δt ≤ 0.001 u/8 h in the case of silicon. This enhanced stability guarantees measurements of isotope ratios on smallest plateaus with lowest uncertainty. The importance of a stable mass scale is pointed out and the relevant quantities of a typical magnetic sector field mass spectrometer are discussed.