Background: Alpha-methylfentanyl and acetyl-alpha-methylfentanyl are two extremely hazardous fentanyl analogues for which addiction data have not been reported. Methods: In this study, we assessed its abuse potential through conditioned position preference (CPP), drug self-administration, and drug discrimination and compared it with fentanyl. Naloxone was also tried in self-administration to suppress relapse behavior. Results: From the CPP results, both fentanyl and alpha-methylfentanyl produced conditioned place preference behavior at a dose of 30 μg/kg, while acetyl-alpha-methylfentanyl produced conditioned place preference behavior at a dose of 900 μg/kg. In self-administration, all three drugs produced a maximum number of infusions at 0.5 μg/kg/infusion, But the number of infusions of acetyl-alpha-methylfentanyl and alpha-methylfentanyl at the peak dose was higher than that of fentanyl. In addition, a single injection of naloxone was also effective in suppressing relapsing behavior in rats. In the drug discrimination experiments, the ED50 of fentanyl, alpha-methylfentanyl and acetyl-alpha-methylfentanyl were 11.06, 12.65 and 128.3 μg/kg, respectively. Discussion and conclusions: By comparing the experimental outcomes, we observed an intriguing phenomenon: the introduction of an alpha-positioned methyl group had a negligible impact on the CPP reward effect and the subjective effect, but significantly reduced reinforcement effect in self-administration experiment. Moreover, the length of the acyl side chain can markedly alter the addiction potential of the drug. Additionally, we tested that a single injection of naloxone can inhibit relapse behavior in rats. In conclusion, evaluation of their structure-addiction relationships can help predict the mechanisms of human drug addiction and identify potential treatment targets.