Double-strand DNA breaks can be repaired by any of several alternative mechanisms that differ greatly in the nature of the final repaired products. We used a reporter construct, designated "Repair reporter 3" (Rr3), to measure the relative usage of these pathways in Drosophila germ cells. The method works by creating a double-strand break at a specific location such that expression of the red fluorescent protein, DsRed, in the next generation can be used to infer the frequency at which each pathway was used. A key feature of this approach is that most data come from phenotypic scoring, thus allowing large sample sizes and considerable precision in measurements. Specifically, we measured the proportion of breaks repaired by (1) conversion repair, (2) nonhomologous end joining (NHEJ), or (3) single-strand annealing (SSA). For conversion repair, the frequency of mitotic crossing over in the germ line indicates the relative prevalence of repair by double Holliday junction (DHJ) formation vs. the synthesis-dependent strand annealing (SDSA) pathway. We used this method to show that breaks occurring early in germ-line development were much more frequently repaired via single-strand annealing and much less likely to be repaired by end joining compared with identical breaks occurring later in development. Conversion repair was relatively rare when breaks were made either very early or very late in development, but was much more frequent in between. Significantly, the changes in relative usage occurred in a compensatory fashion, such that an increase in one pathway was accompanied by decreases in others. This negative correlation is interpreted to mean that the pathways for double-strand break repair compete with each other to handle a given breakage event.