Models for growth in disordered media, which are deterministic once th
e random environment is specified, can be transformed into stochastic
growth processes with a time-dependent noise. The mapping generally in
duces a memory in the stochastic process. This transformation is propo
sed and discussed in detail. An approximation scheme is worked out ana
lytically and possible applications are discussed.