Rates of hospital readmissions are used as indicators of quality of health care. Yet specific causes of readmissions have not been sufficiently studied and an unified definition of such an indicator is lacking. Our goal was to determine the causes and potential avoidability of readmissions in our hospital and to identify a suitable definition of this indicator. Cross sectional study. Hospital Costa del Sol (Marbella, Málaga). Subjects of study: Random sample of hospital discharges followed by a new admission within the next 6 months (n = 363). Determination of the causes and potential avoidability of readmissions, by means of a peer-review of medical records. Descriptive and logistic regression analysis of the variables contained in the Minimum Basic Data Set (MBDS) and related to the avoidability of readmissions. 19% (95% CI: 15.0-23.0) of hospital readmissions within 6 months after hospital discharge were attributed to potentially avoidable situations if the applied medical care had been modified during the previous episode. This figure reached 37% (95% CI: 27.4-47.3) for surgical services, 13% (95% CI: 8.7-17.6) for medical services and 12% (95% CI: 2.9-20.6) for obstetrics. The MBDS variables related to the potential avoidability of readmissions were, for the whole hospital, 1) the shorter interval from the previous discharge and 2) the difference of diagnosis between both episodes. Most hospital readmissions are due to non-avoidable situations if the applied medical care had been modified during the previous episode. In order to detect potentially avoidable situations, definitions of readmission rates should only include early readmissions with a chief diagnosis other than that in the previous admission.