In a single model manual assembly line, product flows through series of workstations arranged in a sequential manner. Each workstation has a finite number of tasks and each task has probabilistic processing time. Due to the probabilistic nature of task time, the task times can exceed the expected standard task time at some instance. If a series of tasks exceeds in a particular station, then there is a risk that the product may exceed the cycle time. As a result, a small variability in task time can lead to large delays in the delivery lead time of the product. Most of the line balancing approaches assume deterministic task times thereby ignoring the impact of task time variability on the system performance measures. The larger the variability of task time, the higher the risk associated with the station. In this paper, the impact of variability in task time is quantified in terms of risk. Risk is defined as potential loss caused when the product fails to complete within the specified station time. For line balancing, in addition to cycle time balancing, the risk should be balanced in order to improve the performance of the assembly line. In this research, a risk based assembly line balancing technique for highly variable task times is presented. The results from the case study show that the method increases the performance of the assembly line while balancing the risk of delays at each station.