This paper discusses the Einstein-Podolsky-Rosen paradox from a new point of view. In section II, the arguments by which Einstein, Podolsky and Rosen reach their paradoxical conclusions are presented. They are found to rest on two critical assumptions: (a) that before a measurement is made on a system consisting of two non-interacting but correlated sub-systems, the state of the entire system is exactly represented by: ψ a (r̄ 1 ,r̄ 2 )=∑ η a η τ η (r̄ 1 ,r̄ 2 (...) )=∑ i,k α ik ψ i (r̄ 1 )σ k (r̄ 2 ) (b) that the exact measurement of an observable A in one of the sub-systems is possible. In section III it is shown that assumption (b) is incorrect. Thus we conclude, as did Bohr, that the results of Einstein, Podolsky and Rosen are not valid. The arguments of section III are quite distinct from Bohr's, and therefore in Section IV this work is related to that of Bohr. (shrink)