Gravel loss is the reduction in thickness of the surfacing course of gravel road. It can be causedrnby many factors. Traffic as average daily traffic (ADT), climate in terms of mean monthlyrnprecipitation (MMP), geometric design parameters of the roads represented by absoluterngradient (G), surface material of the road in terms of plasticity index (PI) and the duration ofrnobservation (D) was used as the model independent variable.rnThe dependent variable of the model, gravel loss, was collected by observing the selectedrngravel roads in three month interval. The selected roads were D/Birhan-Kasima, D/Birhan-rnAnkober and Chacha-Tsgerada â€“Ginager. The samples were prepared in to a grid of 15m ofrnwhich divided into 5m interval. Elevation, across the width of the road, was observed usingrntotal station. Independent variables included in the model were gathered using standardrnprocedure and methods.rnTwo models were developed using SPSS 15.0.The first model used multiple linear regressionrn(MLR) model and it had a prediction power of 87.9% (R2 of 0.879). The second model usedrngeneral linear modeling (GLM) and it had a prediction power of 99.4% (R2 of 0.994).rnThe model helps the parties involved in managing gravel roads. It can be used in gravel roadsrnmanagement system; in determination of maintenance and rehabilitation requirement and inrnpredicting the residual life of gravel roads.