Abstract: On completion of execution of a current block of instructions, a block completion process searches for potential successor blocks, using block descriptors and egress data structures. For each potential successor block, the process compares a set of entry conditions associated with the block with the exit conditions of the current block and, if a match is found, selects the potential successor block as the current block and executes it. A consistency check is performed, to compare the block identity of the successor block with an expected block identity. Block-following code is selectively planted into translated blocks, to call a successor block directly, by-passing the block completion process. The block-following code is optimised, in that it contains tests for entry conditions only if the results of those tests are not known at the time the block-following code is planted.