To better understand the Apennine Front, we have analyzed 14 Apollo 15 impact melt rocks, including hand samples, rake samples, and coarse fines, for major (fused bead microprobe) and trace element (INAA) abundances. The melt rocks show a remarkable diversity of chemical composition, although all but one have typical KREEP incompatible-element patterns. They appear to form five chemical groups, all but the most KREEP-rich of which may represent single impacts. Three groups are of low-K Fra Mauro composition; none correspond to very high alumina (VHA) compositions. The compositions are unlike nearly all other impact melts and regoliths from highland sites, indicating a distinct provenance for the Apollo 15 samples. Serenitatis basin impact melt compositions (Apollo 17 poikilitic boulders) are not duplicated, although one Apollo 17 aphanitic melt sample (73215) is similar to one of the Apollo groups. The Apollo 15 group consisting only of 15445 and 14455 remains the best candidate for Imbrium basin impact melt, while the most KREEP-rich group probably consists of melts produced in varied post-Imbrium impacts into the Apennine Bench Formation (Apollo 15 KREEP basalt flows). The most populated group was previously unrecognized and could be either pre- or post-Imbrium basin; it is possible that it, not 15445/15455, represents Imbrium melt. One unique sample has a non-KREEP incompatible element pattern with low abundances and a positive Eu anomaly. Its target was rather primitive and would not appear to be of regional extent or present in the area in post-Imbrium times; it is either a pre-Imbrian melt or exotic. The Apennine Front does not consist of a monolithic LKFM composition but shows a much wider diversity of compositions than was previously recognized. ¿ American Geophysical Union 1987