This study aimed to develop and validate a green method for the detection of ketamine and its metabolites in human urine samples and subsequently determine which metabolite is more suitable for judgment of ketamine abuse. Ketamine and its metabolites were extracted from urine samples with 1-undecanol using liquid-liquid microextraction based on the solidification of floating organic droplet extraction. The extracts were directly analyzed using gas chromatography-mass spectrometry (GC-MS) without derivatization. The sample pretreatment procedure prior to GC-MS analysis was simple, fast, and required little solvent consumption. Parameters that affect the extraction efficiency, including pH of the urine sample and centrifugation speed, were optimized. Under the optimized conditions, the limits of detection for norketamine, dehydronorketamine, and ketamine were 1.5, 1.7 and 1.0 ng/mL, respectively. The method was used to analyze eight real urine samples from drug abusers and exhibited acceptable accuracy and precision. By analyzing real samples, this study revealed that detection of dehydronorketamine in urine samples for the judgment of ketamine abuse may be more suitable than detection of norketamine or ketamine. The method used in this study addresses the need for green analytical techniques in toxicology.