The cyclin-dependent kinase inhibitors known as p15, p16, p18 and p19 have been suggested as candidates for tumor suppressor genes. The main genetic alterations are deletions (bi- or monoallelic) or 5' CpG island methylation of p15 and p16; very few cases or cell lines had p18 or p19 deletions or hypermethylation. Hypermethylation and homozygous deletions of tumor suppressor genes establish a new paradigm of inactivation by lack of expression, in contrast to the previously identified tumor suppressors which are predominantly inactivated by point mutations followed by loss of the wild-type allele. Here, the literature data on alterations of this gene family in more than 4700 primary cases of leukemia or lymphoma and some 320 continuous leukemia-lymphoma cell lines are summarized. Among hematopoietic malignancies, the highest frequencies of p15del and p16del were seen in acute lymphoblastic leukemia (ALL) (>30%) with striking rates in T-ALL (>50%), but also high rates in B cell precursor (BCP)-ALL (>20%); the rates of deletions in chronic lymphoid leukemia (CLL), multiple myeloma, acute and chronic myeloid leukemia (AML and CML), and myelodysplastic syndromes (MDS) were rather low, only some B cell and T cell lymphomas showed increased frequencies. Results are quite different with regard to the second mode of inactivation, hypermethylation of the promoter region. Here, p15 is most often inactivated, at particularly high frequencies in the disorders lacking any p15/p16 deletions: 40-80% p15met in AML, MDS and multiple myeloma. Also p15met rates in BCP- and T-ALL cases were high (c. 40%). There is controversy concerning the prognostic impact of p15 and p16 aberrations with some studies describing a significant correlation between inactivation of these genes and poor prognosis, while most others did not detect any prognostic relevance, at least in pediatric ALL; there may be a worse prognosis for adults with B or T cell lymphomas. Despite the small number of cases studied, paired sequential analyses suggested that disease progression is associated with loss of p15/p16 activity in a certain percentage of adult patients. p15del/p16del and p15met/p16met were also detected in the large panel of leukemia-lymphoma cell lines studied. In general, the results in cell lines reproduce the data seen in primary cells with the important difference that the rates of p15/p16 inactivation are clearly higher in the cultured cells compared with the freshly explanted cells. Retrovirus- or electroporation-mediated ectopic gene transfer of p16 wild-type into p16-deficient cell lines led to growth inhibition, arrest in G1 (without apoptosis) and occasionally to differentiation, suggesting that the malignant phenotype of p16-/- cell lines can, at least partially, be reversed by restoring p16 gene expression. A striking inverse correlation between the absence of p16 (due to deletion) and presence of wild-type retinoblastoma gene was observed in cell lines confirming a common growth suppressor pathway; no comparable relationship of p16 inactivation with p53 was detected. Paired analysis of cell lines and corresponding primary cell material showed that in all instances tested both populations carried the same gene configuration of p15 and p16. Thus, p15del or p16del did not occur during establishment of the cell lines or during prolonged culture. It is likely that p15 or p16 deletions already acquired in vivo provide a dramatic growth advantage for the immortalization process in vitro, thus increasing the success rate for cell line establishment which is commonly extremely difficult. In conclusion, the present review suggests an involvement of the p15 and p16 tumor suppressor genes in leukemo- and lymphomagenesis. Future studies will determine their exact role in the development and progression of hematopoietic neoplasms. These genes may represent interesting targets for new therapeutic strategies.