The monitoring of MRD by FISH and PCR is now clinically evaluated for a prediction of the relapse of leukemia. However, these methods include several technical limitations. In FISH, an appropriate cut off value should be settled for each probe or procedure in estimating chromosomal gain, loss, or translocation and the sensitivity depends on the cut off values but it may reach to 10(-3). In PCR, the sensitivity is higher than that of FISH especially using genomic DNA as a template. However, the detection of chromosomal translocation-specific DNA or RNA constructs is currently applicable to limited cases and the amplification of Ig or TCR gene rearrangement has a major pitfall caused by the clonal evolution. Clinically, MRD can be an indicator for a prediction of the relapse. For example, a greater MRD on entering CR tends to be related with an early relapse, a return to MRD-positive after disappearance of MRD will be a sign of impending relapse, and MRD negativity at the termination of therapy may be correlated with a long-term disease free status. More precise evaluation of MRD is necessary with regard to therapeutic strategy.