The invention discloses a quadrilateral mutation filtering method for telemetry data, which relates to the technical field of data processing and aims to solve the technical problem that mutation data affects an analysis result of the telemetry data. The filtering method comprises the following steps: 1) acquiring the telemetry data according to a set interval delta T and using data of N points or at least two points to calculate a slope; 2) calculating a quadrilateral trend line slope, namely calculating a data change slope KFi which is equal to (Ki+2Ki-1...+2Ki-n+2+Ki-n+1)/(2n-2) by using an N-point method; 3) drawing a quadrilateral according to the slope, wherein a long side of the quadrilateral is a timeline, while a short side is a floating width W of a trend line; 4) setting coordinate points of a starting edge and an ending edge of the quadrilateral; and 5) detecting mutation data. The method has the characteristics of directly removing the telemetry data with mutation, providing predicted trend data and guaranteeing higher data quality.