We propose a method to estimate the radius, dip and strike of a circular seismic rupture through the inversion of first P- and S-wave pulse widths measured on microearthquake records. The method is based on quite general, numerically calibrated relationships relating source and Q parameters under the assumption that the waves are radiated by a circular crack propagating at a constant rupture velocity. To study the influence of both the source and receiver configuration and the properties of seismic rupture, a detailed resolution study on synthetic data has been carried out. For a microearthquake rupture, the pulse width variations with azimuths depend critically on the fault plane orientation and the resolution on fault angle parameters can be qualitatively assessed by the parameter η, defined as the relative variation of the take-off angles with respect to their average value. A statistical approach based on mapping random deviations on data in the (δ, ø) parameter space has been adopted to quantify the uncertainty affecting the fault plane estimates. The method is applied to estimate the source parameters of an ML= 3.1 event recorded during the 1997 Umbria-Marche earthquake sequence. For the considered event, the fault plane solution is in good agreement with the δ-ø estimates obtained by the method of the joint inversion of P polarities and S polarizations.