At this writing, there is no physical measure that can evaluate consistently the auditory source width (ASW) perceived for sound fields consisting of a direct sound and reflections with various frequency components. In the experiments, a music motif is used as a source signal. The frequency components of reflections are varied by changing the cutoff frequency of low-pass, high-pass, and band-elimination filters, and simultaneously their directions are varied from (plus or minus)18 to (plus or minus)90 deg by steps of 9 deg. ASW and some physical measures are obtained for all sound fields in the experiments. The results on ASW indicate that any frequency component of lateral reflections contributes to create ASW almost equally. The comparisons of ASW with many physical measures show that ICCavg and Lfavg can evaluate ASW for all sound fields consistently with high accuracy. They are the average of the degrees of interaural cross correlation and the lateral energy fractions for 1/1-oct bands for 125 Hz to 8 kHz, respectively. Furthermore, the average of the degrees of cross correlation between two points 30 cm apart without using a dummy head for 1/1-oct bands for 125 Hz to 8 kHz is also adequate, although it is less accurate than ICCavg and Lfavg.