The problem of optimization of throughput and manufacturing operation intensities for a number of time periods for a multi-product production/assembly line under random demand is considered. The objective is to minimize the total cost. The input sequence of products is composed of cyclically repeated identical batches. A batch can be composed of identical or different products. The products of a batch are processed one by one at each position in the order of their arrival. At each time point, only a single product is disposed at each position. The line functioning consists of cyclically repeated sequence of takts, where at each takt operations for all products located at the positions are simultaneously executed. The set of operations to be executed at a takt depends on product and position. The probability distribution of possible demand for each product is assumed known for each time period. For each position, the set of options for equipment which can be used to perform the corresponding sets of operations is given. The batch and selected equipment are invariable over all periods. The cost of each takt includes costs of its operations and an additional cost linearly depending on the takt duration. The sum of manufacturing cost, average costs of storage and/or average penalties for unmet demand on products is used as the objective function. A decomposition method for solving the problem is proposed.