Carbonate clumped isotope thermometry is a powerful new technique increasingly used in many fields in earth science. Recently, it has been shown that clumped isotope measurements can be performed with a Kiel carbonate preparation device and micro-volume analyses, allowing measurements of small (1.5–2 mg) carbonate samples. However, common data correction schemes rely on measurements of gases prepared offline, potentially leading to unrecognized biases in the results. We propose a new correction scheme for the Kiel device method including: (1) A pressure-sensitive baseline correction (PBL) of the raw beam signals; (2) Transfer of data to the absolute reference frame; (3) Correction for acid fractionation; (4) Correction for average standard offsets; (5) When necessary, correction for Δ47 scale compression based on offsets among standards with different ordering state. The long-term performance of the new scheme was tested with a large set of standard measurements (N = 432) obtained over the course of 15 months. The PBL correction reliably removes composition-dependent artifacts, which are commonly corrected for with gas measurements, and offsets observed in micro-volume measurements when ion beams are imbalanced. We show that the shape of the PBL can vary strongly and needs to be properly characterized. Combined PBL and standard correction resulted in long-term stability with standard deviations in Δ47 of 0.012–0.016 ‰ for the five standards over the whole period, close to the average error of 0.011 ‰ observed for individual measurements consisting of 10 replicate analyses. Our correction scheme eliminates the need for routine gas measurements, allowing for equal treatment of samples and standards with the Kiel device setup. While the PBL and standard data obtained over 15 months reveal variable mass spectrometer behavior, they provide a robust means of correction, yielding reproducible results from small carbonate samples in the long term. Copyright