Abstract: There can be lit t le doub t that w e are en tering a new era in ou r understanding of the o rigin s of hum an cancer. U nfo rtunately f rom the po in t of view of the cancer ep idem io logy comm un ity, som e of the mo re recen t advances in the mo lecu lar b io logy of cancer (once fu lly assim ilated) w ill tend to m ake the talk of the up 2to2date cancer ep idem io logist a great deal less st raigh tfo rw ard than m any of u s had p reviou sly envisaged it to be, There m ay st ill be a few cancers that w ill p rove to resu lt f rom on ly a few dist inct ive types of m u tat ion in a relat ively sm allnum ber of genes, bu t I st rongly su spect that the great m ajo rity of hum an cancers that w e w ish to study w ill p rove to have their o rigin s in a comp lex set of DNA changes w ho se p recise m akeup w ill almo st certain ly differ f rom cancer to cancer, individual to individual, and popu lat ion to popu lat ion. I have no doub t that there w ill tu rn ou t to be a relat ively sm all set of genesw ho sem u tan t fo rm s are f requen t ly encoun tered in almo st any type of hum an cancer. Equally, how ever, I su spect that nonm u tan t fo rm s ofm any of the sam e genesw ill tu rn ou t to be impo rtan t in a sim ilarly w ide range of cancers, no t becau se of their w ild2type DNA sequences bu t becau se of a reduct ion (o r increase) in their ab ility to exp ress their encoded funct ion s resu lt ing f rom an ep igenet icmodif i2 cat ion of one fo rm o r ano ther (e. g. by CpG m ethylat ion). Mo re impo rtan t ly, perhap s, is the likelihood that a large m ajo rity of the commoner types of hum an cancer w ill p rove to have their o rigin s in a relat ively large num2 ber of separate genet ic and ep igenet ic changes to the ch romo som alDNA mo lecu les of cells of the o riginat ing t is2 sue. Som e of these changes m ay invo lve m u tato r (o r ep im u tato r) m u tat ion s; th is imp lies that no t all of the changes that lead to a cancer in a m u lt ip ly m u tan t (o r m u lt ip ly ep im u tan t) cellw ill arise independen t ly of each o ther. Sim ilarly, as a tumou r develop s and becom es sub ject to either genet ic o r ep igenet ic in stab ility, it m ayw ell acqu ire num erou s m u tat ion s (o r ep im u tat ion s) w ho se ro le in cau sat ion of the tumo r m ay seem obviou s at f irst sigh t (fo r funct ional reason s) bu t w ho se p resence in that part icu lar tumo r cell has no th ing w hat soever to do w ith the o rigin s of the tumo r. W ith th is background, it seem smo st un likely that the search fo r a single en2 vironm en tal cau se of a given tope of tumo rw ill be info rm at ive un less there are som e very unu sual circum stances w h ich help to en su re that it w ill be. Clearly there are casesw here“b roadscale”ep idem io logy has been ex t rem e2 ly successfu l, fo r examp le w here one studies the relat ion sh ip betw een a comp lex set of environm en tal variab les (cigaret te smok ing o r no t, nat ional dietary hab it s, degrees of sun ligh t expo su re, af latox in expo su re and related liver dam aging circum stances, etc. ) , bu t at temp t ing to ex tend the so rt s of app roaches that w ere successfu l in these cases to the iden t if icat ion of a relat ion sh ip betw een a single compound and a specif ic type of tumo r seem s to m e to be an ex t rem ely op t im ist ic endeavo r. Clearly m any of the m u tat ion s w h ich arise in m u lt ip ly m u tan t cancer cells m u st have their u lt im ate o rigin s in the ho st cell it self, so that m any of the m u tagen s invo lved are genu inely endogenou s to the ho st. Modifying spon taneou sm u tat ion f requenciesm ay therefo re be an impo rtan telem en t in carcinogenesis, and yet it m ay no t be discern ib le in any test system that is cu rren t ly in w idesp read u se. A lthough w e beve know n fo r m uch mo re than a decade that m any of the mo st common fo rm s of hum an cancer are m u lt im u tat ional in o rigin (and to th is w e m u st now add the po ssib ility of m u lt ip le ep im u tat ional changes) , w e seem to have been slow to act on the mo re obviou s imp licat ion s of th is fact. M y su sp icion is that in fu tu re studies of cancer in relat ion sh ip to the environm en t w e w ill need to change f rom a reduct ion ist ap2 p roach ( in w h ich one o r two variab les and a few genes are exam ined) to an app roach in w h ich comp lex ity is f reely acknow ledged and the in teract ion s of m u lt ip le system s are a m ajo r focu s of ou r effo rt s.