The noise impact of the whole railway infrastructure was characterized in the urban environment of Pisa, Italy. The ordinary train transits were considered, nevertheless it was given particular attention also to the noise sources referable to railway operations like manoeuvring, loading and unloading, truck movements, braking, squeals and whistles. These kinds of noise are usually neglected in the noise modelling and are hereafter called "unconventional noises". The characteristics of the railway infrastructure and the receptors' distribution guided the measuring point selection and led to a survey with a sample of 119 people ranging between the ages of 35 and 70 and residents in the area for at least 5 years. The differences between the ordinary noise modelling and the measured noise, including the unconventional ones, were investigated. Dose-effect relationships for %HA and measured or simulated railways noise were calculated and compared with others in literature. The last paragraph of this paper is dedicated to the exposure to railway vibration and its relation with noise exposure. The results show the limitations of traditional noise mapping for railway epidemiological studies based exclusively on ordinary transits and confirm the role of vibrations as enhancing factor for disturbance.