Visualization of mechanical properties of tissue can aid in noninvasive pathology diagnosis. Shear wave elastography (SWE) measures the elastic properties of soft tissues by estimation of local shear wave propagation speed. In this paper, a new robust method for estimation of shear wave speed is introduced which has the potential for simplifying continuous filtering and real-time elasticity processing. Shear waves were generated by external mechanical excitation and imaged at a high frame rate. Three homogeneous phantoms of varying elastic moduli and one inclusion phantom were imaged. Waves propagating in separate directions were filtered and shear wave speed was estimated by inversion of the 1-D first-order wave equation. Final 2-D shear wave speed maps were constructed by weighted averaging of estimates from opposite traveling directions. Shear wave speed results for phantoms with gelatin concentrations of 5%, 7%, and 9% were 1.52 ± 0.10 m/s, 1.86 ± 0.10 m/s, and 2.37 ± 0.15 m/s, respectively, which were consistent with estimates computed from three other conventional methods, as well as compression tests done with a commercial texture analyzer. The method was shown to be able to reconstruct a 2-D speed map of an inclusion phantom with good image quality and variance comparable to conventional methods. Suggestions for further work are given.