This paper proposes a practical algorithm for the reduction of measurement errors due to drift in micro-electro-mechanical system (MEMS) gyros which is used for mobile robot. Drift in MEMS gyros will cause the unbounded growth of errors in the estimation of yaw, which makes it nearly useless in applications that require good accuracy for longer time. The method used in this paper is called "fuzzy heuristic drift reduction" (FHDR). To verify the validity of the algorithm, the paper presents results of experiments, in which a gyro-equipped indoor mobile robot walked for several minutes. FHDR reduced the final heading error over all of these drives by one order of magnitude compared of nonuse.