Reliable estimates of the anelastic attenuation factor, Q, are desirable for improved resolution through inverse Q deconvolution and to facilitate amplitude analysis, Q is a useful petrophysical parameter itself, yet Q is rarely measured. Estimates must currently be made from borehole seismology. This paper presents a simple technique for determining Q from conventional surface seismic common midpoint (CMP) gathers. It is essentially the classic spectral ratio method applied on a trace-by-trace basis to a designatured and NMO stretch-corrected CMP gather. The variation of apparent Q versus offset (QVO) is extrapolated to give a zero-offset Q estimate. Studies on synthetics suggest that, for reasonable data quality (S/N ratios better than 3:1, shallow (<5 degrees ) dips, and stacking velocity accuracy <5%), source-to-reflector average Q is recoverable to within some 3% and Q for a specific interval (depending on its two-way time thickness and depth) is recoverable to 15-20%. Three case studies are reported. First, Q versus offset and vertical seismic profiling (VSP) Q estimates for a southern North Sea line were in close agreement, validating the method. For Chalk, Mushelkalk-Keuper, and Bunter-Zechstein, Q was estimated as 130+ or -15.47+ or -8. and 156+ or -18, respectively. Next, two alternative lithological interpretations of a structure seen in a frontier area were discriminated between when Q estimates of 680 to 820 were obtained (compared to some 130-170 in the overlying units), favoring a metamorphic/crystalline lithology rather than (prospective) sediments. This was later confirmed by drilling. Third, a profile of Q estimates along a 200-ms-thick interval, known to include a gas reservoir, showed a clear and systematic reduction in Q to a low of 50+ or -11, coincident with the maximum reservoir thickness, from some 90-105 outside the reservoir, Q for the reservoir interval itself was estimated at 17+ or -7.