We present the advantages of using partial transmission cord blocks throughout treatment, as opposed to adding full-thickness blocks near the end. Such blocks reduce the risk of block omission or mispositioning and require less total time for construction. We also present an argument for the existence of an optimal width for cord blocks used in mediastinal treatments. A figure of merit has been derived which quantifies the tradeoff between narrow blocks, which increase the variation in dose across the block shadow and the risk of positioning errors, and wide blocks, which may unnecessarily shield potential tumor sites. For 60Co, 4- and 10-MV beams, the figure of merit peaks at block widths of 2.0-2.5 cm at the level of the cord. Effective transmission data for cord blocks constructed of cerrobend are given for those three beams. Quality assurance studies show that transmission through cerrobend blocks can be controlled to the required precision.