Because of the strong interest in the magmatism and volcanism at Long Valley caldera, eastern California, and because of recent significant improvements in our knowledge of the caldera velocity structure and earthquake locations, I have reanalyzed the local-earthquake S-to-P amplitude-ratio data of Sanders (1984) for the gross three-dimensional attenuation structure of the upper 10 km of Long Valley caldera. The primary goals of the analysis are to provide more accurate constraints on the depths of the attenuation anomalies using improved knowledge of the ray locations and an objective inversion procedure. The new image of the high S wave attenuation anomaly in the west-central caldera suggests that the top of the principal anomaly is at 7-km depth, which is 2 km deeper than previously determined. Because of poor resolution in much of the region, some of the data remain unsatisfied by the final attenuation model. This unmodeled data may imply unresolved attenuation anomalies, perhaps small anomalies in the kilometer or two just above the central-caldera anomaly and perhaps a larger anomaly at about 7-km depth in the northwest caldera or somewhere beneath the Mono Craters. The central-caldera S wave attenuation anomaly has a location similar to mapped regions of low teleseismic P wave velocity, crustal inflation, reduced density, and aseismicity, strongly suggesting magmatic association. ¿ American Geophysical Union 1993