A causeofdeathisthemorbidcondition orother disease process, abnormality, orinjury leading directly or indirectly todeath...Fromthestandpoint ofthe prevention ofdeaths itisimportant tocutthechain of events orinstitute thecureatsomepoint. Themost effective public health objective, ofcourse, istoprevent theprecipitating cause fromoperating. Forthis purpose themostuseful single statistic isthatrelating tothe underlying cause ofdeath which maybedefined as(a) the disease orinjury whichinitiated thetrain ofmorbid events leading directly todeath, or(b) thecircumstances oftheaccident orviolence whichproduced thefatal injury ...Symptoms ormodesofdying suchasheart failure, asthenia etc., arenotconsidered tobecauses of death forstatistical purposes. (W.H.O., 1948). Inthevastmajority ofcasesthese elementary rules ofprocedure present little difficulty tothe doctor whohastosign adeath certificate. Mistakes aresometimes made, buteachcertificate iscarefully scrutinized attheGeneral Register Office andif there isanyreason tosuppose that thestated cause of deathmerely describes themodeofdying itis promptly returned foramendment. Inthecaseof stillbirths andneonatal deaths, however, therules cannot bestrictly enforced. Theabsence ofdefinite signs ofdisease (ortheco-existence ofseveral equally lethal conditions), combined withthe difficulty offollowing intra-uterine events, frequently makesitimpossible toexpress areasoned opinion astothecauseofdeath. Itisobvious, forexample, that prematurity cannot possibly have"initiated the trainof eventsleading directly to death". Nevertheless everyyearthousands ofcertificates giving this asthecauseofdeath areaccepted bythe General Register Office, and"prematurity" regularly takes itsplace asacauseofdeath bothinofficial vital statistics andintheinternational list ofdiseases. Atfirst sight certificates whichnamesomething moreprecise, like acongenital defect, asthecause of deathseemmoreinformative thanthosewhich attribute deathtoprematurity. Buta moment's thought will showthat itisthesecondary nature of * Inreceipt ofagrant fromtheMedical Research Council. prematurity andnotanyvagueness intheterm(or anydoubtwhether ababyshould besodescribed) thatmakesitsounrevealing asacauseofdeath. Since maldevelopment, likeunder-development, is theend-result ofanunknownprocess, evencertificateswithsuchprecise diagnostic termsas anencephalus andexomphalos donomorethan describe amodeofdying. Therealization thata medical classification of stillbirths andneonatal deaths isunsatisfactory has ledworkers tostress theimportance ofnon-medical conditions, andtostudy social andgenetic factors which(though theywill never gainadmission tothe International list) arenevertheless ona parwith somecategories nowaccepted ascauses ofdeath (Baird andIllsley, 1953; Douglas, 1950; Karnand Penrose, 1952;Sutherland, 1949; Woolf,1947). Theresults ofsuchinvestigations mayeventually provide uswithamoresatisfactory nomenclature. Meanwhile there issomuchconfusion aboutwhat areproximate andremote causes ofdeath ininfancy, andwhataremerely modesofdying, that itmaybe ofvalue totake agroup oflive anddeadbabies and, without prejudice to thequestion ofcausation, compare thefrequency ofvarious observations made before, during, andimmediately after delivery. Withthis endinviewwehaveexamined theantenatal andobstetrical records ofjust overathousand stillbirths andneonatal deaths, andhavecompared themwiththerecords ofanequal numberofbabies whosurvived beyond theageof4weeks.