This paper tests for price patterns in retail gasoline markets consistent with those predicted by models of implicit collusion among firms. Recent supergame models show that the highest supportable collusive price is a function of today’s profit relative to expected future profit: collusive prices are higher when predictable changes in demand or cost lead firms to expect that collusive profits are increasing rather than declining. Ceteris paribus, collusive profits will be expected to increase when demand is expected to increase and/or costs are expected to decline. Using panel data on sales volume, and retail and wholesale prices in 59 cities over 72 months, we find results consistent with these predictions. Controlling for current demand and input price, the elasticity of current retail margins with respect to expected next-month demand is about 0.38. The elasticity of current margins with respect to next-month wholesale price is about 0.32. The results are inconsistent with inventory effects.