Abstract. This paper deals with a new problem: the Dial and Ride Problem with Split Loads (DARPSL), while using randomized greedy insertion techniques together with constraint propagation techniques. Though it focuses here on the static versions of Dial and Ride, it takes into account the fact that practical DARP has to be handled according to a dynamical point of view, and even, in some case, in real time contexts. So, the kind of algorithmic solution which is proposed here, aim at making easier to bridge both points of view. First, we propose the general framework of the model and discuss the link with dynamical DARP, second, we describe the two algorithms (DARP and DARPSL), and lastly, show numerical experiments for both.