Objective To investigate the protective effect of recombinant adeno-associated virus (rAAV) mediated nerve growth factor (NGF) transfection by intramyocardial injection or subarachnoid injection in type 1 diabetic rats. Methods The whole experiment consists of two sub-experiments, experiment A:12 SPF grade male SD rats were randomly divided into control group (group control) and diabetes mellitus group (group DM, n=6). Group DM was given streptozotocin (STZ, 50 mg/kg, i.p) to induce type 1 diabetic rat model. The tail flick latencies were measured one day before STZ injection and 1, 2, 4, 6, 8 and 9 weeks after STZ injection in two groups of rats. The contents of NGF and calcitonin gene-related peptide (CGRP) were measured by ELISA method. Experiment B:24 SPF grade male SD rats were randomly divided into 4 group (n=6): diabetic heart transfection control group (group MC), diabetic heart transfection group (group ME), diabetic spinal cord transfection control group (group SC) and diabetic spinal cord transfection group (group SE). The 2×2 factorial design was used in this study. The diabetic rat model was established by giving STZ (50 mg/kg, i.p) in four groups of rats. At the end of 4 weeks after injection of STZ, rats in group MC and ME were transfected with rAAV-green florescent protein (GFP) and rAAV-NGF-GFP at a titer of 0.8×1013 virus genomes (μg/L) by intramyocardial injection at 100 μL. Rats in group SC and SE were transfected with rAAV-GFP and rAAV-NGF-GFP at a titer of 0.8 × 1012 virus genomes (μg / L) by subarachnoid injection at 25 μL. After 5 weeks, the cardiac function indexes were measured in groups of rats. The expressions of GFP in myocardium, spinal cord and dorsal root ganglion of T1-T5 were observed by fluorescence microscope, and the contents of NGF and CGRP were measured by ELISA method. Results In experiment A, after 4 weeks, the tail flick latency was significantly delayed in group DM compared with that of group control (P<0.05). The NGF and CGRP protein in myocardium were significantly decreased in group DM compared with those of group control (P<0.05). In experiment B, both rAAV-NGF-GFP and myocardial tissue as transfection pathway significantly improved the cardiac function of rats (P<0.05), and up-regulated the contents of NGF and CGRP protein in myocardial tissue (P<0.05). There was an interaction between transfectants and transfection pathways during the experiment. Combination of both can improve cardiac function and up-regulate NGF and CGRP in myocardium (P<0.05). The myocardial point injection of rAAV-NGFGFP can significantly improve the cardiac function of rats (P<0.01), and up-regulate the contents of NGF and CGRP protein in myocardial tissue (P<0.01). Conclusion Compared with subarachnoid injection, intramyocardial injection of rAAV-NGF can effectively increase the expression of NGF in myocardial tissue of diabetic rats and have a more effective cardioprotective effect. eng [ABSTRACT FROM AUTHOR]