We have developed a method that simultaneously determines focal mechanism solutions of many small earthquakes and source-region station corrections for short-period body-wave amplitudes by inverting amplitude data of P, SH, and SV waves, together with P-wave polarity data. The observed seismic waveform includes the effects of site amplification and attenuation along its ray path in addition to the radiation pattern of earthquake source. The amplitude of seismograms at frequencies higher than a few Hertz is extremely sensitive to heterogeneous structure near the ground surface. Consequently, we need to know in detail the effects of site amplification and attenuation in order to estimate focal mechanisms by using short-period waveforms. However, at present, we do not know the detailed crustal structure with a resolution necessary for this estimation. In the present study, we assume that P- and S-wave attenuation factors along ray paths from hypocenters to each station can be expressed as a function of hypocentral distance, backazimuth, and incident angle. Based on this assumption, we determined focal mechanism solutions of many earthquakes and the coefficients in the function for each station simultaneously, by using P-, SH-, and SV-wave amplitudes and P-wave polarities. We applied the present method to 170 aftershocks of the 1996 Onikobe earthquake (M 5.9), which occurred in the central part of northeastern Japan. We obtained focal mechanism solutions of many microearthquakes whose mechanism solutions could not be determined by using P-wave polarity data alone. P axes of almost all the obtained focal mechanism solutions are horizontal and oriented in the east-west direction. T axes are, on average, near vertical at the shallowest depth. As the depth approaches 5 km, the T axes become horizontal and then gradually become near vertical again.