The vehicle routing problem is one of the most practical issues in the field of operation research that has been addressed since the 60s, and lots of efforts have been made by researchers in this field, which has led to the great advances. The vehicle routing problem is one of the most well-known optimization problems, with the aim of designing the optimal set of routes for serving customers in a manner that is consistent with existing constraints. Different types of problems which address the servicing issue are seen in real world and the physical delivery of goods is the most common types of them. In general, these problems include a warehouse, a fleet of vehicles set up in the depot, and a set of customers that should be served. In the simplest case the goal of this problem is to assign the vehicles so that overall cost of the tours will be minimized based on maximum capacity of the vehicles and also the maximum working times. In this research, the vehicle routing problem with a three-dimensional loading constraint and also when the split delivery of demands is allowed has been investigated. The three-dimensional loading implies that customers requesting rectangular cube items. For this reason, weight, length, width and height of the items should be specified. The goal is to find the optimal allocation of customers to vehicles, so that the total cost of service is minimized and the loading of items assigned to each vehicle will be feasible. The split delivery of demands implies that each customer can be served by vehicles more than once, therefore, unlike some kinds of vehicle routing problems, which all the requested items of a customer must deliver in a vehicle, in this research, with the split delivery of demands, you can place customer-requested items in multiple vehicles. For solving the model, Simulated Annealing algorithm and genetic algorithm have been used. In order to validate these methods, several problems have been solved using both methods and have been compared by solving the GAMS software. The results of their comparison and analysis clearly show that the methods used are acceptable methods for solving the model. Key words: vehicle routing problem, three-dimensional loading, split delivery of demands, meta-heuristics algorithms.