Restrictionfragment length polymorphisms (RFLPs) have been proposed for predicting the yield potential of different types of cross. Their has been evaluated in 15 flint and 12 dent inbreds from the European maize {Zea mays L.) germplasm as well as in 68 F, crosses (21 flint X flint, 14 dent X dent, and 33 flint X dent) produced between them. The materials were evaluated for Fi performance and midparent heterosis of grain yield, dry matter content (DMC) and plant height in two environments in Germany. Genetic distances (GDs) between parental lines, calculated from RFLP data of 194 cloneenzyme combinations, showed greater means for flint X dent (0.67) than dent X dent (0.62) and flint X flint (0.55) crosses. Cluster analysis based on GDs resulted in a clear separation of flint and dent hnes and agreed well with pedigree information. For the complete set of 64 crosses analyzed (excluding 4 crosses between closely related lines) correlations of GD with Fi performance and heterosis were significant for all traits except F] performance of DMC. When separate calculations were performed for individual subjets of crosses, correlations of GD with Fi performance and heterosis were significantly positive (0.48 < r < 0.80) for all traits in the flint x flint crosses, but not significant for the subsets of flint x dent and dent x dent crosses. Our results confirm those of previous investigations in that the predictive value of RFLP data is restricted to crosses between lines from the same heterotic group, and cannot be applied to crosses between lines from genetically divergent heterotic groups.Recognition of inbred lines with superior performance in crosses is a key for the success of hybrid breeding programme. Traditional methods have relied primarily on extensive field evaluation of crosses in topcross of diallel tests and are very time-consuming and costly. The efficiency of hybrid breeding programme would be improved if superior crosses could be predicted reliably (before field tests) by the simple screening of inbred lines.Parental genetic distance, determined from assays of inbreds for isozymes or restriction fragment length polymorphisms (RFLPs), has been suggested as a potential tool for predicting the hybrid performance of crosses. This approach has been stimulated by quantitativegenetic theory and empirical studies indicating that the amount of heterosis in crosses is, within limits, positively related to the parental genetic divergence (see HALLAUER and MIRAN-U.S.