"The relation implies that a $10 increase in the price of a barrel of Brent crude oil is typically associated with a 25 cent increase in the average U.S. retail price of a gallon of gasoline. The relation only captures the long-run tendency, and leaves out seasonal factors that for example brought the price of gasoline temporarily lower this last winter. But since this spring U.S. gasoline prices have moved back in line with what you’d expect given the long-run fundamentals."
(There is a nice gasoline price estimator based on current barrel of brent crude at that post.)
However, Dave Giles (with a nice econometrics blog) recently linked to a paper he authored (PDF) testing whether movements up and down in the price of gasoline are symmetrical-- he found they are not. Giles finds gasoline prices respond more rapidly to oil price increases than decreases. Giles is using data from Vancouver, other studies using different countries' data have confirmed the so-called "rockets and feathers hypothesis."
The purported possibilities for this phenomenon, as listed by Giles, are interesting. On the more highly-competitive end, think of firms' prices asymptotically approaching cost (where their profits are then $0). When one firm suddenly raises its price, the competitor can follow and the asymptotic movement toward cost begins again. Another possibility is that consumers are less choosy when they see everyone's prices falling; that puts less competitive pressure on all firms to lower their prices. Both are interesting hypotheses. I suspect my behavior is akin to the latter; I'm less likely to look for a "better deal" when I see all prices are low relative to what they were last week/month than I am when I see a relatively high price at one station I drive by.