Performing two tasks at the same time involves the risk of between-task interference, measured as the backward crosstalk effect (BCE). The BCE is larger the more simultaneous the tasks appear, i.e., with shorter stimulus-onset asynchronies (SOA). Fischer and Dreisbach (2015) found that stimuli associatively linked with short SOA facilitated task shielding of Task 1 (i.e., smaller BCE). Röttger et al. (2021) tested whether implicit knowledge about SOA sequences across several blocks would facilitate task shielding as well. Using a dual-task paradigm with spatial response overlap (manual-pedal left-right responses) they did not find differences in the BCE between blocks of fixed or random SOA sequences, however. In the present online study, we adapted the design of Röttger et al. (2021) and used a dual task with identical task sets to maximize dimensional overlap. A simplified sequence of six SOAs (short-short-medium-medium-long-long) was used for four blocks with a subsequent block of random SOA presentation. Results showed that the size of the BCE remained constant throughout the sequence blocks but increased for the random block, suggesting that implicit knowledge about SOA sequences was used for task shielding. These preliminary findings will be replicated in a second laboratory study that adds another sequential block after the random block.