The increasing popularity of screen content in mainstream video applications promotes its continuous development. These applications generate a great capacity of data and bring an enormous burden to existing limited network bandwidth and storage space. Rate control (RC) is a key coding technology to improve video compression efficiency. However, existing RC methods ignore the subjective perceptual characteristics of screen content. Thus, we propose a subjective bit allocation scheme for screen content coding, making full use of the subjective perception characteristics. Firstly, the rate-distortion (R-D) characteristics of screen content coding are statistically analyze, and a novel frame-level subjective R-D model is built. Then, a frame-level bit allocation method is derived based on the R-D model and on the Lagrange optimal method. Finally, experimental results show that proposed method can save up to 7.98% bitrate against the baseline method under similar subjective quality.