PIP: Accurate classification of perinatal deaths is instrumental in efforts to improve the perinatal mortality rate (PNMR). Accurate records should be maintained on birth weight, gestational age, a simple classification of cause of death (possibly the "Wigglesworth" or the author's system), a classification system correlated with birth weight/gestation of dead and live births, and, where possible, detailed coding such as the International Classification of Diseases (ICD) or the Systematized Nomenclature of Medicine (SNOMED). The fetal loss rate in Great Britain is reported as a minimum of 0.1% per week for pregnancies from 22 weeks until term. Obstetric or neonatal management can affect the level of mortality. Terminations due to malformations performed before the 24th week are found to account for a 50% decline in the PNMR. PNMR is affected also by birth weight and the occurrence of multiple births. Neonatal care that prolongs the survival of sick or malformed neonates may decrease PNMR or increase infant death rates. Analysis of trends in PNMR must consider demographic changes such as an increase in preterm multiple births due to fertility treatment. Perinatal deaths may be classified by cause with a variety of systems. The ICD system is a single axis classification which uses three digit codes for primary causes and a fourth and fifth digit for detailed subcategories, which can be accessed in a alphabetically-organized index. SNOMED provides six different fields of information, including topography, morphology, aetiology, function, disease, and procedure. Five subordinate fields are identified and linked to topographic body sites. Choice of classification systems is considered to be based on availability of information and the needs for subsequent analysis. Two other systems of classification are available in the UK. The Aberdeen system is based on maternal factors, but the disadvantage is the high number of unexplained deaths. The Hey et al. system codes by fetal and neonatal factors and does not require modern investigative techniques, but the detail is highly variables from case-to-case and hospital-to-hospital. The author's system is based on routine autopsies of external conditions. Four groupings are identified, including macerated normally-formed stillborn infants, congenital anomalies (stillbirth or neonatal death), conditions associated with prematurity, and fresh stillbirths presumed asphyxiated. A fifth category groups infants dying of specific conditions such as toxoplasmosis.