Abstract Seismic signal analysis can provide valuable information to understand rockslide dynamics in the absence of direct observation. Established methods of moment tensor inversion may assist with rapid rockslide parameter estimation, which can be crucial for mitigating secondary hazards. In contrast to previous studies that focused only on large rock avalanches, we inverted regional long-period seismic data from two medium-sized rockslides (volumes around 150,000 m 3 ), which occurred in 2005 and 2009 near the Swiss border in the central Alps. The resulting deviatoric and full moment tensors were found to be persistent using different station combinations, and the goodness of fit is high for both events (∼70%). In contrast, we found that the long-period signals are not well represented by a single-force source. Differences observed in the full moment tensor components for the two rockslides may reflect their different source processes, whereas the orientations of the double-couple/compensated linear vector dipole axes are roughly aligned with source area normal vectors. We used a moment tensor grid-search method to locate both rockslide events, because the goodness of fit decreases with increasing distance to the true locations. The best estimated locations were situated 6.3 and 6.4 km from the true event locations for the two rockslides. Our results encourage further analysis of seismic waveforms to help determine the source parameters of rockslides (location and mechanism) at regional distances. With adequate velocity models, station geometry, and broadband seismic data, automated algorithms could enable rapid rockslide detection and characterization.