Unplanned medical 30 day readmissions place a burden on the provision of acute hospital services and are increasingly used as quality indicators to assess quality of care in hospitals. Multivariable logistic regression of a 10 year database showed that four factors were most strongly associated with early readmission: Charlson comorbidity index >=1, respiratory disease as a principal diagnosis, liver disease and alcohol-related illness as an additional diagnosis, and the number of previous readmissions. Disease and patient-related factors beyond control of the hospital are the factors most strongly associated with 30 day readmission to hospital, suggesting that this may not be an appropriate quality indicator.