Disassembly Line Balancing Problem considered here both chooses the best disassembly alternative for an end of life product and assigns the corresponding disassembly tasks to the workstations of the line with the aim to reduce the line cost. Precedence and cycle time constraints are observed. Task times are assumed stochastic with known normal probability distributions. The line cost includes the investment and operation costs for workstations as well as penalty costs generated by the cycle time constraints violations. To deal with uncertainties, a stochastic linear mixed integer formulation is developed. To solve efficiently the problem, L-shaped algorithm combined with Latin Hypercube Sampling is proposed.