Existing counterfactual data augmentation methods for sequential recommendation only consider users’ implicit feedback for generating augmented counterfactual samples, while the explicit feedback is ignored. Therefore, we propose an Explicit and Implicit Counterfactual data Augmentation algorithm for Sequential Recommendation (EI-CASR) to address this issue. Our algorithm takes into account both explicit and implicit feedback information of users. By learning the logical inverse (NOT) operation, neural logical reasoning can model explicit feedback in sequential learning, thus making it possible to conduct counterfactual reasoning over users’ explicit feedback to generate explicit counterfactual samples for data augmentation. At the same time, the implicit sampler generates implicit counterfactual samples by replacing historical items of user interaction. Two sets of augmented training samples, together with the original training samples, can help improve the recommendation performance by generating synthetic data to cover the unexplored input space. Experimental results on three public datasets demonstrate that EI-CASR significantly improves the performance of sequential recommendation tasks and effectively addresses the data sparsity problem commonly encountered in sequential recommendations.