This study was aimed at determining how a simple experimental control—recalling the sequence of digits in the order they were presented—may affect quantitative and qualitative performance in Sternberg's short-term memory scanning paradigm. First, the presence or absence of this constraint was shown to have practically no effect on the relationship between sequence length and mean response time: the two general laws proposed by Sternberg were found to hold true in both experimental conditions. However, subjects who had to recall the digit sequence were consistently slower than those who did not, and the yes response curves of the two experimental conditions differed in shape. Moreover, correlations between parameters of the linear function and various aptitude scores showed that this task was linked to the memory span only when the participants had to recall the digit sequence. If our results are confirmed, they should help to improve the interpretation of the data collected in many studies that use this paradigm in cognitive psychology as well as in neuroscience of memory.