Since horizontal curves on rural highways are among the most challenging geometric elements, this study aimed to estimate the CMF associated with increasing the curve radius on rural highway horizontal curves in Iran. To cover diverse regions of the country, seven provinces — Tehran, Isfahan, Fars, Mazandaran, Khorasan-Razavi, Qazvin, and Kermanshah — were selected.Due to data limitations in crash records and intervention histories, a cross-sectional study method was adopted for CMF estimation. Crash data (fatal and injury crashes) from 2019 to 2021 were obtained from the Road Safety and Traffic Office of the Road Maintenance and Transportation Organization. Using GIS software, the spatial distribution of crashes along highway routes — with a focus on horizontal curves — in the selected provinces was mapped. To obtain data on traffic volume and road length required for cross-sectional modeling, ArcGIS tools were utilized to calculate the length of curve segments. Traffic counter data (including counter locations, traffic volumes, and the percentage of heavy vehicles) were extracted from the website of the Road Maintenance and Transportation Organization (site 141). Then, in QGIS software, traffic volumes were assigned to each road segment based on the spatial position of traffic counters. To enhance model accuracy, additional information on curve improvement variables — such as curve radius, median type, and overtaking control conditions — was collected from provincial road safety and traffic offices. The negative binomial regression model, commonly used in crash modeling literature, was employed to develop the CMF using SAS software. Modeling results showed that the CMF for increasing the curve radius was 0.49, indicating a 51% reduction in crashes. Model validation, conducted using RMSE, MAE, Mean Residual, and Pseudo R² statistics, confirmed the good accuracy of the model, suggesting that the estimated CMF values are reliable.