The potential contribution of abnormal marrow stromal function to ineffective haemopoiesis in the myelodysplastic syndromes is unclear. We have compared the ability of stromal layers from normal (n = 7) and myelodysplastic (n = 9) marrow to alter proliferation and survival of the granulocyte-macrophage colony-stimulating factor/interleukin-3-dependent cell line F-36P. Co-cultures for 72 h in the absence of exogenous cytokines were either in direct contact with stroma or separated by transwell inserts. On normal stromal layers, the ratio of adherent F-36P cells relative to stromal cells increased from a mean of 0.2 ± 0.01 (s.d.) at 4 h of co-culture to 0.34 ± 0.08 after 72 h (n = 7). Corresponding values on myelodysplastic stroma (0.2 ± 0.02 at 4 h and 0.35 ± 0.05 at 72 h; n = 9) indicated that the ability of myelodysplastic stromal layers to regulate short-term proliferation of F-36P cells may be similar to normal. Apoptosis of F-36P cells was quantified after co-culture with normal or myelodysplastic stroma: results from myelodysplastic co-cultures were standardized as a fraction of values from co-cultures with paired normal stroma (apoptotic ratio). Augmented apoptosis of F-36P cells was detected in 8/9 co-cultures with myelodysplastic stroma (mean = 15.7 ± 9.7%, n = 9), compared with corresponding normal stroma (mean = 12.4 ± 4.6%, n = 7, P < 0.05) with a mean apoptotic ratio of 1.4 ± 0.5 (P < 0.05). There was no correlation between stroma-related apoptosis and FAB type, tumour necrosis factor- concentrations in the culture supernatant or numbers of stromal macrophages, and no evidence of involvement of the Fas pathway. Increased apoptosis was detected in cells grown in transwell inserts over stroma (23.8 ± 3%, n = 5) compared to adherent cells in cultures with normal stromal layers, but this survival difference was not observed in co-cultures with myelodysplastic stroma. These results suggest that abnormal stromal function in patients with myelodysplastic syndromes may contribute to increased apoptosis of haemopoietic cells within the marrow microenvironment. The effect appears to be dependent on close cellular contact, rather than the release of soluble factors, but the exact mechanism remains unclear.