The invention relates to an intelligent transformer station online monitoring parameter trend early warning method. According to the method, the historical monitoring data of an online monitoring device for a time segment after stabilization is taken as samples, a relevant coefficient of a monitoring value and the time is calculated, and an early warning threshold is determined according to the relevant coefficient; before early warning analysis, multiple samples are continuously acquired from the historical monitoring data, a relevant coefficient of a monitoring value and the time for each sample is calculated, the relevant coefficients and the early warning threshold are compared, if the relevant coefficient trends to continuously grow and is greater than the early warning threshold, an early warning signal is outputted. The method is advantaged in that trend analysis and early warning can be automatically carried out, trend analysis accuracy is improved, and workload of equipment maintenance personnel is reduced.