The retinal degeneration slow (rds or Prph2^Rd2/Rd2) mouse, a model of recessive retinitis pigmentosa, lacks a functional gene encoding peripherin 2. This membrane glycoprotein is required for the formation of photoreceptor outer segment discs. The striking feature of the rds mouse is the complete failure to develop outer segments. We have previously examined the short-term effect of gene replacement therapy using an adeno-associated (AAV) vector and demonstrated induction of outer segments and improvement of photoreceptor function. Here we have extended our analysis and have demonstrated that the potential for ultrastructural improvement is dependent upon the age at which animals are treated, but the effect of a single injection on photoreceptor ultrastructure may be long-term. However, there was no significant effect on photoreceptor cell loss, irrespective of the date of administration, despite the improvements in morphology and function. Our investigation excluded procedure-related damage, vector toxicity and immune responses as major factors which might counteract the benefits of photoreceptor restoration, but suggested that transgene over-expression is of significance. These findings suggest that successful gene therapy in patients with photoreceptor defects may ultimately depend upon intervention in early stages of disease and upon accurate control of transgene expression.