The rapid rise in import competition in recent years provides an unusual test for examining the market structure-performance relationship with panel data. A weakness of the cross-sectional approach is that many important variables are difficult to measure. However, if these variables are determined by slowly evolving technological conditions and if there is time-series variation in an observable variable such as import shares, then the use of a "within-industry" estimator may control for some of the problems confronting cross-section studies. Our estimates indicate a sizable effect of import competition on price-cost margins for the time period 1976-1986.