A computerized version of the Corsi blocks task (Milner, 1971) was assessed for standard forward-recall order (Experiments 1 and 3) and for reversed-recall order (Experiments 2 and 3) either in a single-task or in a dual-task design combined with articulatory suppression, matrix-tapping, random-interval generation or fixed-interval generation as concurrent tasks during the encoding stage. Concurrent performance of the matrix-tapping task impaired memory performance for short as well as for longer block sequences. The random-interval generation task, which loads executive processes, impaired memory performance mainly at intermediate- and longer-sequence lengths, while fixed-interval generation, which is presumed to put no load on executive processing, did not show any effect. Articulatory suppression did not impair memory performance on forward-recall order, but it impaired memory for longer sequences in the backward-recall condition in Experiment 2, but not in Experiment 3. The results are discussed within the context of the working-memory model of Baddeley and Hitch (1974).