The relation between fidelity of implementation and student outcomes in a computer-based middle school mathematics curriculum was measured empirically. Participants included 485 students and 23 teachers from 11 public middle schools across seven states. Implementation fidelity was defined using two constructs: fidelity to structure and fidelity to process. Because of the nested nature of the data, we used a two-level hierarchical linear model for analysis. Four variables, all categorized as fidelity to structure variables, proved significant--total time in intervention (p less than 0.001), concentration of time in intervention (p = 0.03), direct observation of intervention fidelity (p = 0.04), and pretest score (p less than 0.001). Fidelity to process was found to be nonsignificant. The importance of measuring the relation between implementation fidelity and student outcomes is discussed as well as implications for researchers and teachers. (Contains 5 tables.)