Recently, multi-scenario learning (MSL) has achieved flourishing development in recommendation systems of E-commerce platforms. Current numerous models have been proposed that attempt to use a unified model to serve multiple scenarios. In these works, user’s sequential behavior modeling is infrastructure for capturing the dynamic nature of the user profiles and ensuring their predictive performance. However, recent multi-scenario learning methods usually ignore the utilization of item attributes when modeling behavior, which contain crucial prior information for deciding which items to recommend. Specifically, considering item IDs only in the model will lead to insufficient characterization of interests especially in highly sparse recommender settings. Meanwhile, the lack of abundant attribute information is not conducive to model the significant interests discrepancy among multiple scenarios. In this paper, we address these limitations by proposing a Scenario and Attribute-aware Contrastive Network (SACN) for multi-scenario learning. First, SACN employs Item-level Preference Extracting (IPE) and Attribute-level Preference Extracting (APE) modules capturing user’s coarse-grained and fine-grained preferences dynamically in terms of scenario context and item attributes. Then, by introducing self-supervised learning in Scenario Contrastive Module (SCM), we can distinguish users’ interests in different scenarios sufficiently and achieve superior performance. Offline experiments on real-world datasets and online A/B testing both show that the proposed model significantly outperforms all state-of-the-art models in the task of multi-scenario recommendation (MSR).