In neonatal rats, superior colliculus (SC) ablation results in a massive and rapid increase in retinal ganglion cell (RGC) death that peaks about 24 h post-lesion (PL). Naturally occurring cell death during normal development, and RGC death after axonal injury in neonatal and adult rats, has primarily been ascribed to apoptosis. Given that normal developmental cell death is reported to involve caspase 3 activation, and blocking caspase activity in adults reduces axotomy-induced death, we examined whether blocking caspases in vivo reduces RGC death after neonatal SC lesions. Neither general nor specific caspase inhibitors increased neonatal RGC survival 6 and 24 h PL. These inhibitors were, however, effective in blocking caspases in another well-defined in vitro apoptosis model, the corpus luteum. Caspase 3 protein and mRNA levels in retinas from normal and SC-lesioned neonatal rats were assessed 3, 6 and 24 h after SC removal using immunohistochemistry, western and northern blots and quantitative real-time polymerase chain reaction. Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labelling (TUNEL) was used to independently monitor retinal cell death. The polymerase chain reaction data showed a small but insignificant increase in caspase 3 mRNA in retinas 24 h PL. Western blot analysis did not reveal a significant shift to cleaved (activated) caspase 3 protein. There was a small increase in the number of cleaved caspase 3 immunolabelled cells in the ganglion cell layer 24 h PL but this represented only a fraction of the death revealed by TUNEL. Together, these data indicate that, unlike the situation in adults, most lesion-induced RGC death in neonatal rats occurs independently of caspase activation.