The number of fatal crashes along the Federal Road FT050 (Jalan Batu Pahat - Kluang - Ayer Hitam) involving students of Universiti Tun Hussein Onn Malaysia (UTHM) are still accelerating. Road quality upgrading such as lanes expansion, upgrading road surface and construction of road divider are carried out to improve road and traffic safety, since 2010. Unfortunately, the total number of crashes did not reduce but even showed a tendency to increase. Therefore, the objectives of this study are to analyses the crash trends, determine the rank of hazardous road section and identifying the impact of the road divider in an attempt to reduce road fatalities. This study is focused on the road along FT050, which start from the Kolej Vokasional Batu Pahat to the Kolej Jururawat Parit Raja (KM7 - KM21 or 14 km lengths) where the road divider was constructed. Crash data have been collected from Polis Diraja Malaysia such as crashes data before the road divider are constructed (2010 - 2012) and after the construction (2013 – 2014) and road geometry design data are collected from Jabatan Kerja Raya for road design assessments. The determination of Accident Point Weightage (APW) was performed by applying the APW method and Microsoft Excel, which was used for hazardous road ranking. The Accident Prediction Model developed based on multiple linear regression are used to relate the crashes rates and the road design parameters. The results show sufficient proof to support the road safety relate to the relationship between the vehicles crash numbers and road design and the number of crashes namely accident point weighting is described as the dependent variable. The following results found the existing number of major access points, at the following signalized and unsignalized intersections, number of traffic volumes/Annual Average Daily Traffic (AADT) and road geometry are the main potential contributors of increment crash rates on multiple lane rural roadways. Keywords: Crash analysis, black spot ranking, road divider, crash prediction model