Abstract: At present, seismic amplitude and frequency attributes, such as the sweetness attribute, are usually used for hydrocarbon detection by most post-stack methods.However, the hydrocarbon detection results based on the common sweetness attribute, are highly ambiguous due to the influences of thickness, interference of adjacent strata, and other non-fluid factors.In this paper, a new method for hydrocarbon detection is proposed that involves fusing a 90° phase shift attribute and its instantaneous frequency attribute.Firstly, the 90° phase shift and instantaneous frequency attributes were extracted based on original zero-phase seismic data.Then, the seismic attributes of hydrocarbon and water reservoirs in drilled wells were used as constraints to optimally solve the fusion parameters.Finally, the amplitude and frequency attributes were fused to obtain the hydrocarbon detection results.Forward modeling results showed that the new method was less affected by the thickness of the reservoir within the range 8~25 m, than the conventional sweetness attribute.The proposed method suppressed the interference of adjacent reservoirs effectively, and was therefore more applicable for fluid detection.The practical application in the Bohai PL oilfield demonstrated that the proposed method successfully distinguished between oil layers and water layers.Furthermore, the spatial distribution of oil-bearing anomalies was in accordance with the law of hydrocarbon accumulation.