The present paper describes the methodological optimisation and validation of a capillary zone electrophoresis method for the determination of morphine, cocaine and 3,4-methylenedioxymethamphetamine (MDMA) in hair, with injection based on field-amplified sample stacking. Diode array UV absorption detection was used to improve analytical selectivity and identification power. Analytical conditions: running buffer 100 mM potassium phosphate adjusted to pH 2.5 with phosphoric acid, applied potential 10 kV, temperature 20 degrees C, injection by electromigration at 10 kV for 10 s, detection by UV absorption at the fixed wavelength of 200 nm or by recording the full spectrum between 190 and 400 nm. Injection conditions: the dried hair extracts were reconstituted with a low-conductivity solvent (0.1 mM formic acid), the injection end of the capillary was dipped in water for 5 s without applying pressure (external rinse step), then a plug of 0.1 mM phosphoric acid was loaded by applying 0.5 psi for 10 s and, finally, the sample was injected electrokinetically at 10 kV for 10 s. Under the described conditions, the limit of detection was 2 ng/ml for MDMA, 8 ng/ml for cocaine and 6 ng/ml for morphine (with a signal-to-noise ratio of 5). The lowest concentration suitable for recording interpretable spectra was about 10-20-times the limit of detection of each analyte. The intraday and day-to-day reproducibility of migration times (n = 6), with internal standardisation, was characterised by R.S.D. values < or = 0.6%; peak area R.S.D.s were better than 10% in intraday and than 15% in day-to-day experiments. Analytical linearity was good with R2 better than 0.9990 for all the analytes.