The invention discloses a baseline identification method for a fetal heart rate curve. A device for implementing the method comprises a signal input detecting module, a baseline identification extracting module, a baseline value parameter calculating module and a baseline value parameter output module. The method comprises the following steps: first, inputting fetal heart rate signals, performing pre-calculation, average optimization and interpolation smoothing in the baseline identification extracting module through pre-treatment and extraction of the effective signals, and calculating a baseline value; then, calculating the baseline proportion in a range and descriptive statistical parameters in a parameter calculating module; finally, outputting the calculating result in the baseline value parameter output module. The baseline value is calculated according to fetal heart rate data of long-term fetal heart rate monitoring and the definition of a fetal heart rate baseline, and has a relatively specific physiological significance; through combination of accelerating, decelerating and variance classifications, counts and duration proportions of the fetal heart rate, more comprehensive information can be obtained from the fetal heart rate signals; through application of the extracted baseline valve, the frequency domain of the fetal heart rate and related entropy information can also be obtained.