BackgroundRepetitive transcranial magnetic stimulation (rTMS) is effective for treatment‐resistant depression (TRD). Optimal rTMS parameters remain unclear, especially whether number of sessions or amount of pulses contribute more to treatment outcome. We hypothesize that treatment outcome depends on the number of sessions rather than on the amount of pulses.MethodsWe searched databases for randomized clinical trials (RCTs) on high‐frequent (HF) or low‐frequent (LF)‐rTMS targeting the left or right DLPFC for TRD. Treatment efficacy was measured using standardized mean difference (SMD), calculated from pre‐ and post‐treatment depression scores. Meta‐regressions were used to explore linear associations between SMD and rTMS pulses, pulses/session and sessions for HF and LF‐rTMS, separately for active and sham‐rTMS. If these variables showed no linear association with SMD, we divided the data into quartiles and explored subgroup SMDs.ResultsEighty‐seven RCTs were included: 67 studied HF‐rTMS, eleven studied LF‐rTMS, and nine studied both. No linear association was found between SMD and amount of pulses or pulses/session for HF and LF‐rTMS. Subgroup analyses showed the largest SMDs for 1200–1500 HF‐pulses/session and 360–450 LF‐pulses/session. The number of sessions was significantly associated with SMD for active HF (β = 0.09, p < 0.05) and LF‐rTMS (β = 0.06, p < 0.01). Thirty was the maximal number of sessions, in the included RCTs.ConclusionMore rTMS sessions, but not more pulses, were associated with improved treatment outcome, in both HF and LF‐rTMS. Our findings suggest that 1200–1500 HF‐pulses/session and 360–450 LF‐pulses/session are already sufficient, and that a treatment course should consist of least 30 sessions for higher chance of response.