Effects of ex vivo GDNF gene delivery on the degeneration of motoneurons were studied in the G1H transgenic mouse model of familial ALS carrying a human superoxide dismutase (SOD1) with a Gly93Ala mutation (Gurney et al., 1994). Retroviral vectors were made to produce human GDNF or E. coli beta-galactosidase (beta-Gal) by transient transfection of the Phoenix cell line and used to infect primary mouse myoblasts. In 6-week-old G1H mice, 50,000 myoblasts per muscle were injected bilaterally into two hindlimb muscles. Untreated G1H and wild-type mice served as additional controls. At 17 weeks of age, 1 week before sacrifice, these muscles were injected with fluorogold (FG) to retrogradely label spinal motoneurons that maintained axonal projections to the muscles. There were significantly more large FG-labeled alpha motoneurons at 18 weeks in GDNF-treated G1H mice than in untreated and beta-Gal-treated G1H mice. A morphometric study of motoneuron size distribution showed that GDNF shifted the size distribution of motoneurons toward larger cells compared with control G1H mice, although the average size and number of large motoneurons in GDNF-treated mice were less than that in wild-type mice. GDNF also prolonged the onset of disease, delayed the deterioration of performance in tests of motor behavior, and slowed muscle atrophy. Quantitative, real-time RT-PCR and PCR showed persistence of transgene mRNA and DNA in muscle for up to 12 weeks postgrafting. These observations demonstrate that ex vivo GDNF gene therapy in a mouse model of FALS promotes the survival of functional motoneurons, suggesting that a similar approach might delay the progression of neurodegeneration in ALS.