In nine healthy humans we modulated corticospinal excitability by using conditioning-test paired-pulse transcranial magnetic stimulation in a repetitive mode (rTMS), and we compared its effect to conventional single-pulse rTMS. We applied 80 single pulses or 80 paired pulses to the motor cortex at frequencies ranging from 0.17 to 5 Hz. The conditioning-test intervals were 2, 5, or 10 ms. Motor evoked potential (MEP) amplitudes from the abductor digiti minimi (ADM) as target muscle and extensor carpi radialis (ECR) indicated the excitability changes during and after rTMS. During paired-pulse rTMS at a facilitatory conditioning-test interval of 10 ms, we observed a facilitation of MEPs at 1, 2, and 5 Hz. A similar facilitation was found during single-pulse rTMS, when stimulus intensity was adjusted to evoke MEPs of comparable size. Using an inhibitory conditioning-test interval of 2 ms, paired-pulse rTMS at frequencies of 1 and 2 Hz caused no change in MEP size during the train. However, paired-pulse rTMS at 5 Hz caused a strong enhancement of MEP size, indicating a loss of paired-pulse inhibition during the rTMS train. Since no facilitatory effect was observed during single-pulse rTMS with an adjusted stimulus intensity, the MEP enhancement during 5 Hz rTMS was specific for "inhibitory" paired-pulse rTMS. After 5 Hz rTMS MEPs were facilitated for 1 min, and this effect was not substantially different between paired-pulse rTMS and single-pulse rTMS. The correlation between ADM and ECR was most pronounced at 5 Hz rTMS. We conclude that paired-pulse rTMS is a suitable tool to study changes in corticospinal excitability during the course of rTMS. In addition, our data suggest that short trains of paired-pulse rTMS are not superior to single-pulse rTMS in inducing lasting inhibition or facilitation.