An effective condition monitoring is essential to increase the availability of Pelton turbines and to avoid unexpected damage. The main challenge consists in determining the deformations and the stresses in the runner while in operation. To do so, it is essential to understand the dynamic response of the machine, and especially of the runner. Moreover, the monitoring locations have to be selected as to detect this response optimally. Having a deep knowledge of the natural modes of the runner, and how these can be excited while it is operating is mandatory to detect damage. The modal behavior of a Pelton runner can either be studied by means of numerical simulation or by experimental modal analysis (impact testing). However, none of these methods alone is able to describe with accuracy the dynamic behavior of the machine under real operating conditions. In this study, an experimental investigation of an existing Pelton turbine unit has been carried out. The modal response of the runner suspended, attached to the shaft and with the machine in operation has been studied. To do so, accelerometers, acoustic emission sensors and other types of sensors have been used. Signals were acquired during transients and with the machine operating at different loads. For a better understanding, a numerical model based on Finite Element Method (FEM) was created to represent the behavior of this type of turbines. At the end, the natural frequencies detected with the machine in operation were compared to those obtained numerically and by impact testing. Therefore, the effect of different factors on the runner response was analyzed. In addition, the best sensor and optimal location in describing the state of the machine were chosen. From the results obtained a possible improvement of the condition monitoring for these turbines is presented.