To get more insight into the phenotypic changes of childhood acute lymphoblastic leukemia (ALL) at relapse, a detailed morphological and immunophenotypic study in 40 childhood ALL cases (32 precursor B-ALL and 8 T-ALL) was performed. Expression patterns of non-lineage specific markers (terminal deoxynucleotidyl transferase (TdT), CD34, and HLA-DR), B-lineage markers (CD10, CD19, CD20, and CD22), T-lineage markers (CD1, CD2, CD3, CD4, CD5, CD7, and CD8), and cross-lineage myeloid markers (CD14, CD15, and CD33) were compared at diagnosis and relapse. In case of low blast counts (< or = 70%) at relapse, double labeling for membrane markers and TdT was used in order to define the precise immunophenotype of the TdT+ leukemic cells. An immunological marker-shift was defined as either a conversion from positive to negative and vice versa or a difference in positivity of > or = 50%. Morphological differences between diagnosis and relapse were detected in 34% of precursor B-ALL and 14% of T-ALL. Differences in immunological marker expression were found in 72% of precursor B-ALL and in 75% of T-ALL, and generally concerned minor shifts with loss or acquisition of a few markers. The morphological shifts and immunophenotypic shifts were not correlated. Immunophenotypic shifts were found for all markers tested in precursor B-ALL, except for HLA-DR. Shifts in CD10 expression (16% of cases) were only observed in relapses occurring 30 months or more after diagnosis. In four precursor B-ALL an intra-lineage shift was found at relapse (one common ALL to null ALL and three pre-B-ALL to common ALL or null ALL) and two precursor B-ALL cases were diagnosed as acute non-lymphocytic leukemia at relapse based on morphology and immunophenotype. In T-ALL, neither intra-lineage nor inter-lineage shifts were observed, although shifts were detected in all T cell markers tested, except for the lineage specific CD3 and T cell receptor (TcR) markers. In conclusion, immunophenotypic shifts at relapse frequently occur in precursor B-ALL and T-ALL, in a small percentage leading to an intra-lineage shift (10%) or inter-lineage shift (5%). Therefore immunophenotypic monitoring of minimal residual disease in ALL patients should be based on multiple marker combinations, preferably together with polymerase chain reaction analysis of rearranged immunoglobulin and/or TcR genes or chromosome aberrations.