Aberrations affecting the tumor suppressor gene p16INK4a have been described for a variety of tumors. In breast cancer, approximately 50% of tumors show low or lack p16 expression. While evidence provided by some studies has implicated a possible role for p16 in normal replicative senescence, other studies have suggested that the Rb, pathway through which p16 functions, may not be involved in senescence control. Previously we observed that all immortal lines derived from normal mammary epithelium which were analysed for p16 displayed inactivation of this gene through distinct mechanisms, supporting p16 inactivation as a possible necessary event in escape from senescence. To further clarify this issue, we have analysed p16 expression in a panel of normal finite lifespan human mammary epithelial cells (HMEC) from initial propagation through growth arrest, using media which confer different replicative capacity. Approximately 10-25-fold increase in p16 expression was observed for all normal HMEC with initial onset of a senescence phenotype following 15-25 population doublings in culture. These cells also displayed expression of the senescence associated beta-galactosidase. Interestingly, HMEC with additional long term replicative capacity (approximately 80 population doublings) arose from these growth arrested cultures, showing lack of p16 expression. This extended growth capacity appears to be associated with a methylation phenomenon since treatment of these cells with the methylation inhibitor 5-aza-2-deoxycytidine resulted in growth arrest concurrent with reacquisition of p16 expression and senescence associated beta-galactosidase. Analysis of p21waf1 expression revealed no change in expression during growth in vitro. These results support p16INK4a as the 9p senescence gene and suggest a role for p16 loss in the escape from initial onset of senescence and in acquisition of an extended life span of human mammary epithelial cells.