Sequential recommendation aims to predict users’ next interactions based on their historical behavior sequences, typically assuming that observed actions reflect user interest. However, user behavior is also influenced by popularity. Most existing methods have not sufficiently disentangled these two factors, limiting their ability to accurately model user preferences. To address this issue, we propose DIPRec (Disentangled Interest and Popularity Modeling with Causal Intervention for Sequential Recommendation) framework, which explicitly models users’ interest preference and popularity preference through a dual-branch architecture. To mitigate the impact of potential confounders in preference estimation, we adopt a causal perspective and introduce front-door adjustment. Moreover, we incorporate time-aware popularity into the modeling to better capture popularity-driven behavior and enhance the disentanglement of the two preference types. In addition, an adaptive fusion module is designed to dynamically balance the influence of interest and popularity based on contextual information. Extensive experiments on three real-world datasets demonstrate the superiority of DIPRec over state-of-the-art baselines.