In 2005 a vertical 42 MW hydropower unit was upgraded in Sweden. One of the requirements was that the dynamic behaviour of the machine should not be affected. Resonance problems became apparent after the hydropower unit was re-commissioned and, in order to remedy this, new measurements and calculations were undertaken. While measuring, the machine went into resonance twice. During normal operation the shaft displacement showed high amplitudes at a frequency of twice the machine's nominal speed. At resonance the displacement amplitude increased and a dominant frequency occurred at ~2.4x nominal speed. To explain the reason for the resonances, new rotor dynamic calculations were performed using non-isotropic bearing and bracket properties. This model showed good correlation between calculated and measured values for the lowest eigenfrequencies. Probable explanations for the resonance are shape deviation in the generator, misalignment of the shaft and low damping of some critical eigenfrequencies close to twice the nominal speed.