This paper presents an experimental comparison of five methods of identifying friction in robot drives. The methods considered are direct plotting of velocity versus armature voltage, plotting velocity versus armature current, third harmonic estimation, batch least squares and sequential least squares. These methods were implemented on a d.c. servo motor robot drive system to identify Coulomb and viscous friction parameters. It is shown that an asymmetric Coulomb and viscous model properly identifies the frictional torque due to the combined sliding and rolling friction in the motor. Furthermore, although each of the identification methods is shown to be capable of giving reasonable estimates of the frictional coefficients, the plotting of velocity versus armature current is shown to be most suitable for off-line frictional identification and the sequential least-squares method most suitable for on-line identification, particularly when coefficients may change with time.