Hospital activity data can be accessed from a variety of sources ranging from hospitals to the Department of Health. These data provide valuable and widely used information, but care is needed in their use and interpretation. Hospital activity rates reflect not only the underlying prevalence and severity of disease, individual factors and referral practices, but also variations related to provider-specific factors: the 'provider effect'. This includes completeness in the data, supply of hospital beds, admission policies, hospital access and distance from hospital. The provider effect can be controlled to a certain extent in statistical analyses. Although data quality has improved considerably in the last decade, this should still be investigated where trusts are being compared and in small area studies because missing data may lead to artefactual differences in rates. 'Dump' postcodes, where missing or unknown postcodes are assigned to a local postcode such as that of the hospital, may affect small area analyses and linkage if a proxy patient identifier is constructed that includes postcode.