Harmine (HAR) and harmaline (HAL) were metabolized by demethylation to form harmol (HOL) and harmalol (HAM) both in vivo and in vitro. It has been demonstrated tremendous value of HAR, HAL and their metabolites in the therapy of Alzheimer's disease. A rapid, selective and sensitive UPLC–ESI-MS/MS method was firstly developed and validated for the simultaneous determination of HAR, HAL, HOL, and HAM in beagle dog plasma with 9-aminoacridine as the internal standard (IS). After protein precipitation with acetonitrile, the analytes were separated within 4.5 min on an ACQUITY UPLC BEH C18 column with a gradient elution system composed of 0.1% formic acid and acetonitrile at a flow rate of 0.4 ml/min. Detection was performed using multiple reactions monitoring mode under a positive ionization condition. The calibration curves of four analytes showed good linearity (r² > 0.9959) within the tested concentration ranges. The low limit of quantification for HAR, HAL, HOL, and HAM were all 1.00 ng/ml. The mean accuracy of the analytes was within the range of 94.56–112.23%, the R.S.D. values of intra-day and the inter-day precision were less than 6.26% and 7.51%, respectively. Matrix effects and extraction recoveries of the analytes from the beagle dog plasma were within the range of 94.48–105.77% and 89.07–101.44%, respectively. The validated method was successfully applied to a pharmacokinetic study of HAR, HAL, HOL, and HAM in beagle dogs after intravenous administration of HAR and HAL both of 1.0 mg/kg. The main pharmacokinetic parameters of Cmax, Vd, CL, AUC and MRT, except Ke and t_1/2 values, showed significant difference between the two parent drug HAR and HAL, respectively (p < 0.05–0.001). Because of the different metabolic rate of HAR and HAL in vivo, the two metabolites, HOL and HAM, exhibited unique pharmacokinetic properties.