Attentional control – also called executive functions or cognitive control – refers to our ability to maintain a goal and goal-relevant information in the face distraction. Recent research has put forward the difficulty of establishing attentional control at the latent-variable level. This questions the measures used so far, thus asking for other ways of measuring attentional control. Early research has modeled attentional control from working-memory tasks and short-term memory tasks (i.e., tasks used to measure either the temporary maintenance and manipulation of information or the temporary maintenance of information only). First, attentional control was modeled as the residual variance of working memory after controlling for short-term memory. Second, it was modeled as the common variance across working-memory and short-term memory measures, with higher loadings of the working-memory measures to the factor of attentional control. The purpose of the present study was to test the replicability and robustness of both models. After a systematic search, we found three datasets with relevant correlation matrices. We estimated the models using the original correlation matrices and 5000 correlation matrices bootstrapped from the original correlation matrices. Across the datasets, the models were not fully and exactly replicated. They required modifications to provide an acceptable fit to the data. Moreover, all these model estimations were not robustly observed when the 5000 simulated correlation matrices were used. Therefore, using working-memory and short-term memory tasks does not solve the difficulty of establishing attentional control as a latent variable. These findings challenge the conceptualization of attentional control.