This paper presents a two-echelon inventory-routing problem for perishable products. Products are delivered from a supplier to an intermediary depot, where storage may occur and from which they are delivered by smaller vehicles to the customer locations. Holding costs are incurred for storage at the depot. Customer availability is taken into account in the form of customer delivery patterns. The objective is to minimise the total transportation and holding costs. We formulate the problem as a mixed integer linear program and solve it by means of an adaptive large neighbourhood search metaheuristic in combination with the solution of a reduced formulation. Three variants of the heuristic are compared on a variety of randomly generated instances. Given the two-stage structure of the problem, computational results show the importance of taking the cost structure into account when choosing the most suitable solution approach.