AbstractResolving robust source parameters of small‐moderate magnitude earthquakes is still a challenge in seismology. We infer directivity from apparent source time functions (ASTFs) at regional distance and quantify the associated uncertainties. ASTFs are used for (i) modeling a propagating 1‐D line source from the duration data and (ii) inverting the 2‐D slip distribution from the full signals. Slip inversion is performed through a Popperian scheme, where random trial models are either falsified on account of large misfit, or else become members of the solution set of the inverse problem. We assess the resolution of rupture directivity representing centroid shifts from the solution set in a rose diagram. Using as example an event with well‐studied rupture directivity, the 2011 Mw 5.2 Lorca (Spain) earthquake, 1‐D and 2‐D parameterizations yield similar estimates for direction (N213°E and N220°E, respectively) and asymmetry (67:33, 65:35) of rupture propagation, as well as rupture length (2.1 km, 2.7 km) and speed (3.5 km/s, 3.25 km/s). The high rupture velocity ≥ 90% vS may be held primarily responsible for the strong directivity effect of this earthquake. We show that inversion of apparent source durations is intrinsically unable to resolve highly asymmetric bilateral ruptures, while inversion of full ASTFs misses part of the signal's complexity, suggesting the presence of deconvolution artifacts. We extend the analysis to the Mw 4.6 foreshock of the Lorca earthquake, inferring similar directivity parameters and slip pattern as for the mainshock. The rupture toward SW of both earthquakes suggests that this direction could be inherent to the fault segment.