The marathon course on road surfaces varies frequently in alignment horizontally and /or vertically. This fact compelled use to use a calibrated bicycle method for the course measurement, and to be required statistical approaches for data processing. This paper deals with the computation of the Seoul Olympic Marathon course lengths measured on May 18, 1986. The concept of safety factor against short course is analyzed for certification, and statistical methods are presented to compute an appropriate safety factor. The results of computation show that the best (lowest) actual measurement provides about 99.95% confidence that the combined lengths of all intervals will not be found short upon an equally accurate remeasurement.