This paper discusses two very relevant practical issues in the application of vibration-based health monitoring to civil engineering structures: the excitation source and the effect of temperature. The idea of vibration-based damage detection is to measure dynamic characteristics such as eigenfrequencies, damping ratios and mode shapes on a regular basis. The state, and eventually degradation, of the structure is reflected in the evolution of these characteristics. Unfortunately, it is not only the health of a structure that influences its measurable dynamics, but also the applied excitation and the changing temperature are important factors and may erode the damage detection potential. In the first part, the results of different excitation types are compared: band-limited noise generated by shakers, an impact from a drop weight and ambient sources such as wind and traffic. In the second part, the undeniable effect of temperature on measured eigenfrequencies is demonstrated and a methodology is proposed to distinguish these temperature effects from rear. damage events. The method could be validated on a unique data set from a bridge that was artificially damaged after a one-year monitoring period.