The purpose was to study the validity of a recently proposed method [Forsell C, Halvorsen K. A method for determining minimal sets of markers for the estimation of center of mass, linear and angular momentum. Journal of Biomechanics 2009;42(3):361-5] for estimating the trajectory of the whole-body center of mass (CoM) in the case of running at velocities ranging from 10 to 22 km h(-1). The method gives an approximation to the CoM using the position of fewer markers on the body than the standard method of tracking each segment of the body. Fourteen male athletes participated. A standard method for determining the CoM from a model of 13 segments and using the position of 36 markers was used as reference method. Leave-one-out cross-validation revealed errors that decreased with increasing number of markers used in the approximative method. Starting from four markers, the error in absolute position of the CoM decreased from 15 mm to 3 mm in each direction. For the velocity of the CoM the estimation bias was neglectable, and the random error decreased from 0.15 to 0.05 m s(-1). The inter-subject and intra-subject variability in the estimated model parameters increased with increasing number of markers. The method worked well also when applied to running at velocities outside the range of velocities in the data used to determine the model parameters. The results indicate that a model using 10 markers represents a good trade-off between simplicity and accuracy, but users must take into account requirements of their specific applications.