Repairable spare parts inventory optimization deals with special type of inventories that are repaired and returned tornusable conditions rather than discarding. Surplus repairable spare parts lead to a high carrying cost, whereasrninadequate spare parts can result into aircraft down time. This is always a problem and it is the trade- off betweenrnthe size of spare parts inventory and carrying cost.rnThere have been numerous studies performed concerning repairable spare parts optimization. However, literaturesrnthat consider no fault found (NFF) and scrap rate (S) as an input variable for repairable inventory systems is, to thernbest of the authorâ€™s knowledge, lacking. Because the existing literatures assume negligible no fault found andrnnegligible scrap rate which actually deny the reality encountered by maintenance organizations in practice. rnThis study deals with repairable spare parts optimization in Ethiopian Maintenance repair and Overhaul division. Inrnorder to reach that goal, multi stage purposive sampling technique including cluster sampling and sequentialrnsampling to determine sample size of repairable spares. Poison distribution, diminishing Marginal Returns andrnmixed integer linear programming (MILP) were applied to four different scenarios to study the impact ofrnmaintenance turnaround time (TAT), no fault found (NFF) and scrap rate on repairable inventory size. Finally,rnmixed integer linear programming (MILP) model developed that minimizes overall inventory costs and improvesrnservice level. rnThe quality assurance of all the data inputs were verified by using sensitivity analysis and overall comparison ofrnmodel out puts. Sensitivity analysis is by transposition of mixed integer linear programming equations and byrnvarying service level. Model outputs were validated by comparison of the model output with the actual existingrnsystem and the results of other research findings.rnThe finding of the study shows better cost saving, compared to the current actual cost. Mixed integer linearrnprogramming model gives largest aggregate inventory cost saving result of 19.57% by considering Scrap Rate andrnNo fault found. Sensitivity analysis indicates that total inventory cost saving of 4% can be achieved by reducing thernshop repair cycle time by 20% without affecting no fault found impact. Combined total inventory saving of 37% canrnbe achieved by reducing the repair cycle time by 20% and by reducing no fault found by 10% without affecting thernservice level. Financial Impact of No fault found (NFF) is about 5 % (4.936%) when compared to baseline scenariornbased on mixed integer linear programming (MILP). The Combined Financial Impact of No fault found (NFF) andrnScrap Rate (SR) for marginal analysis and mixed integer programming models are 13.483% and 29.066% higherrnrespectively when compared to baseline scenario. Based on the cost saving result and the financial impact of NFFrnand SR, it is concluded that mixed integer linear programming (MILP) results superior cost saving as compared torndiminished marginal returns and poisson distribution under consideration of NFF and SR as input variable.