Let A,B be two rings, ρ a homomorphism from A to B and p a prime ideal of A. For there to exist a prime ideal p′ of B such that ρ−1(p′)=p, it is necessary and sufficient that ρ−1(Bρ(p))=p.
If there exists an ideal p′ of B such that ρˉ−1(p′)=p, then ρ(p)⊂p′, whence Bρ(p)⊂p′ and ρˉ−1(Bρ(p))⊂ρˉ−1(p′)⊂p; as the converse inclusion is obvious, σˉ−1(Bρ(p))=p. Conversely, suppose that ρˉ−1(Bρ(p))=p and consider the multiplicative subset S=ρ(A−p) of B; the hypothesis shows that
S∩Bρ(p)=∅;
by Corollary 2 there exists a prime ideal p′ of B containing Bρ(p) and not meeting S; then ρˉ−1(p′) contains p and cannot contain any element of A−p and hence is equal to p.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.