An orthogonal array design (OAD) was applied to optimize microwave-assisted derivatization (MAD) for analysis of trace amphetamine (AM) and methamphetamine (MA) by negative chemical ionization gas chromatography-mass spectrometry (NCI GC-MS). The 2,3,4,5,6-pentafluorobenzoyl chloride (PFBC) was used as a derivatization reagent. Experimental factors including solvent, microwave power, and irradiation time at four-levels were studied in 16 trials by OAD(16) (4(4)). The significance of these factors was investigated using analysis of variance (ANOVA) and percent contribution (PC). Solvent is statistically demonstrated a chief factor; microwave power and irradiation time are secondary factors. Under the optimum condition, calibration curve of AM is linear over a range from 0.01 to 100 ng mL(-1) with correlation coefficient 0.9988, and MA from 0.1 to 1000 ng mL(-1) with correlation coefficient 0.9951. The limit of detection (LOD) is 1.20 pg mL(-1) for AM and 13.04 pg mL(-1) for MA. An applicability of the method was tested by analyzing urine samples from amphetamine-type stimulants (ATS)-abusing suspects. Consequently, the OAD method not only optimizes the MAD condition for determination of trace AM and MA, but identifies the effects of factor solvent, microwave power and irradiation time on the MAD performance.