Lost circulation is one of the most frequent and hazardous complications in drilling operations, and its accurate prediction plays a vital role in ensuring the safe and efficient exploration and development of hydrocarbon resources. However, conventional prediction methods largely rely on historical drilling and logging data combined with empirical analyses, while often neglecting critical geological risk elements such as structural features. These methods suffer from delayed predictions and limited spatial applicability, making them insufficient for pre-drilling risk assessment in complex geological environments. To address these challenges, this study proposes a seismic-guided prediction method for geological lost circulation risks based on a “four-element hazard” (FEH) model. Utilizing multi-scale data from representative offshore drilling blocks, the method integrates well-logging data, drilling parameters, and 3D seismic information. Through geological statistics and analysis of typical well sections, four major geological factors are identified as the primary triggers for lost circulation: fault zones, volcanic conduits, lithologic discontinuities, and abnormally overpressured formations. These factors form the foundation of the FEH model framework. Guided primarily by seismic data and constrained by well and drilling information, the method extracts multi-source sensitive seismic attributes to establish identification workflows for each of the four risk types. Specifically, a multi-attribute Bayesian fusion model is used to estimate fault-related risk probabilities; joint amplitude-variance analysis delineates the boundaries of volcanic conduits; lithologic interface indicators are optimized based on response features; and abnormal overpressure zones are predicted by integrating seismic velocity and pore pressure inversion. Field applications in the Bohai A Block and South China Sea B Block demonstrate strong consistency between predicted risk zones and actual lost circulation events. In particular, the model successfully forecasted 80% of the loss intervals in the Bohai H1 well, including several composite-origin zones, with a maximum instantaneous loss rate of 90 m³/h. These results validate the model’s capability for forward-looking and effective risk prediction in structurally complex formations. In summary, this research develops a three-dimensional, seismic-guided, pre-drilling risk identification workflow targeting structurally complex zones. The method provides essential technical support for well placement optimization, trajectory design, and proactive drilling risk management.