Background: The lumbar sympathetic ganglia are variable in both position and numbers. The aim of this study was to detect the appropriate lumbar vertebral level where the lumbar sympathetic ganglia primarily aggregate. Methods: Forty patients comprising of hyperhidrosis (26 cases), complex regional pain syndrome (10 cases), peripheral artery occlusion disease (3 cases) and postherpetic neuralgia (1 case) underwent lumbar sympathetic block. We randomly selected one of two (L2 or L3) levels and the L4 level. The position of the needle tip and distribution of dye was verified by injection of a mixture of radio-opaque dye (1.5 ml) and 4% lidocaine (1.5 ml) and subsequently confirmed by L-spine anteroposterior and a lateral view X-ray. We considered the response positive when the skin temperature increased more than $1^{circ}C$ in 5 min. Results: In general, the positive response ratio was greater when the needle tip located at the L2 or L4 level vice L3 and when the drug was distributed on the lower half of the L4 body and in the L4/5 intervertebral disc space. In a right side block, the positive response ratio was greater when the drug was distributed on the lower half of the L4 body and in the L4/5 intervertebral disc space, although in a left side block there was no significant difference seen between the levels. The complications of the neurolytic block were alcohol neuritis (7.5%) and hypoesthesia (5%) on the L1 or L2 dermatome. Conclusions: The best effect with least chance of complication may be induced by spreading the drug on the lower half of the L4 body and/or into the L4/5 intervertebral disc space by placing the needle tip on the L4 body.