Conversion of one P-derived transposon into another has already been shown to occur with a measurable frequency. However, the mechanism responsible for such replacements has remained controversial. We previously proposed a mechanism involving three partners. We assumed that after excision of the P-element inserted at the target site, the double-strand break was repaired using, first, the homologous P sequences on the sister chromatid, and second, a remote template, the donor P-derived transposon. However, two other mechanisms have been proposed. The first involves two partners only, the broken end and the remote template, while the second involves transposition of the donor into the target P-element, followed by a double recombination event. Here we describe the conversion of a defective P-element using as a remote template an enhancer-trap element that is itself unable to transpose because it lacks 21 bp at its 5' end. This result makes it possible to exclude the possibility that this conversion event occurred after transposition. The new allele was molecularly and genetically characterized. The occurrence of a polymorphism at position 33 of the P-element sequence and of an imperfect copy of the template on the 3' side of the converted transposon confirmed that the sister chromatid was absolutely necessary as a partner for repair. Our results show that targeting of a marked P-element is possible, even when this element is unable to transpose. This provides a means of improving recovery of conversion events by eliminating unwanted transpositions catalyzed by the P transposase.