Purpose: To determine the optimal formulation of lipid prodrug, 1-O-hexadecyloxypropyl-phospho-ganciclovir (HDP-P-GCV), for intravitreal delivery. Methods: Equal concentrations of crystalline or liposomal HDP-P-GCV were exposed to rabbit whole vitreous, core vitreous, peripheral vitreous, human plasma, and heat inactivated rabbit vitreous, and the samples were incubated at 37degreesC for one week. Aliquots were taken at day 1, 2, 3, and 7 and subjected to HPLC analysis for conversion to GCV. Results: The resultant concentration of GCV from crystalline HDP-P-GCV in vitreous was 198 +/- 49 muM (n = 3) at day 1 and 1253 +/- 248 muM (n = 3) at day 7. The resultant concentration of GCV from the liposomal formulation of HDP-P-GCV in vitreous was much lower, yielding a concentration of 66 +/- 7 muM (n = 3) at day 1 and 243 +/- 39 muM (n = 3) at day 7 (P < 0.001, t Test). When the crystalline HDP-P-GCV was incubated with heat-inactivated vitreous, the detectable GCV concentrations were low (22 μM) and did not increase over time. The concentration of GCV detected from the crystalline HDP-P-GCV in the core vitreous was 19.69 +/- 3.84 μM (n = 3) at day 1 and 1537.36 +/- 177.14 μM (n = 3) at day 7. The concentration of GCV released from crystalline HDP-P-GCV in peripheral vitreous was 32.86 +/- 5.07 μM (n = 3) at day 1 and 1805.78 +/- 327.94 μM (n = 3) at day 7. Detectable GCV concentration from both core and peripheral vitreous samples increased over time, however, the magnitude of GCV release from peripheral vitreous samples was higher (P < 0.05, t Test). Conclusion: In vitreous, HDP-P-GCV as a crystalline formulation was converted to GCV more rapidly than liposomal formulation of HDP-P-GCV. Vitreous cells may play an important role in the metabolism of either formulation of HDP-P-GCV delivered into vitreous.