Epiretinal deposits of triamcinolone acetonide (TA) can be detrimental to retinal cells in vitro as several laboratory studies have shown. This contrasts with the good clinical experience of intravitreal TA use. We investigated the effect of TA crystals on retinal cells concerning the critical dose range, a potential cell recovery, the drug-tissue interaction and what protective biological factors could explain the discrepancy between in vivo and in vitro results. A human retinal pigment epithelium cell line (ARPE19) and transformed rat retinal ganglion cells (RGC5) were used. Purified TA crystals were either added directly on top of the cell cultures or on top of membrane filter inserts, basement membrane sheets or porcine vitreous with the cells growing underneath. To determine the number of live versus dead cells fluorescent stains were used. Proliferation and viability were measured using the MTT assay and the mean inhibitory dose (ID(50)) calculated with or without a filter. Cell recovery was measured after transient TA exposure (0.01-1 mg/ml) compared to continuous exposure after 7 days. To exclude a mere mechanical effect of epicellular deposits the TA crystals were replaced by glass pearls in a serum-free medium and the MTT toxicity assay was performed after 24 h. Without direct contact of TA crystals with the cells only a moderate decrease of mitochondrial activity was observed that fully recovered after transient exposure and showed a clinically safe ID(50) of 7.7 mg/ml. In contrast, direct exposure to even minute crystalline deposits for 7 days caused a rapid progressive and irreversible cell death being significant far below clinically used concentrations (ID(50) 0.058 mg/ml). Direct exposure to glass pearls did not show any loss of viability. Both basement membrane sheets and vitreous reliably prevented direct cytotoxicity to underlying retinal ganglion cells. Our findings suggest that irreversible TA cytotoxicity in a cell culture setting occurs earlier than previously assumed in the presence of even minute epicellular deposits. But in most clinical situations epiretinal TA deposits seem not to be harmful to ocular cells as protective biological factors may prevent close apposition of TA crystals to susceptible retinal cells. However, in eyes that have undergone vitrectomy with ILM peeling epimacular deposits could be critical.