Students' attitudes toward mathematics and the strength of their mathematics preparation typically go hand in hand such that their specific effects are difficult to disentangle. Employing the method of propensity weighting of a continuous variable, we built hierarchical linear models in which mathematics attitudes and preparation are uncorrelated. Data used came from a national survey of U.S. college students taking introductory calculus (N = 5,676). A 1-standard-deviation increase in mathematics preparation predicted a 4.72-point higher college calculus grade, whereas a 1-standard-deviation increase in mathematics attitudes resulted in a 3.15-point gain. Thus, the effect of mathematics preparation was about 1.5 times that of mathematics attitudes. The two variables did not interact, nor was there any interaction between gender and these variables.