We examined the safety of acetylcholine (ACh) and ergonovine (ER) tests retrospectively and investigated the optimal protocol of provocation test for diagnosis of multivessel coronary spasm. We performed 1546 ACh tests and 1114 ER tests during 23 years. ACh was injected in incremental doses of 20/50/80 μg into the right coronary artery (RCA) and of 20/50/100/200 μg into the left coronary artery (LCA) over 20 s. ER was administered in total doses of 40 μg into the RCA and of 64 μg into the LCA over 2–4 min. When a coronary spasm was induced and did not resolve spontaneously within 3 min after the completion of ACh/ER injection, or when hemodynamic instability due to coronary spasms occurred, 2.5–5.0 mg of nitrate was administered into the responsible vessel. To relive provoked spasm, it is necessary to administer nitrate in 31 cases by ACh and in 76 cases by ER (2.0 vs. 6.8 %, p < 0.01) before another vessel attempts. Multivessel spasms were often observed in LCA testing than in RCA testing on both agents [ACh: 78.6 % (11/14) vs. 11.8 % (2/17), p < 0.001, ER: 37.8 % (14/37) vs. 20.5 % (8/39), ns]. Even after the administration of nitrates, positive coronary spasm was obtained in 21.1 % by ACh and 52.9 % by ER tests on another coronary artery. No irreversible complications were recognized on both tests. We should firstly perform spasm provocation tests in the LCA and we may be able to diagnose another vessel spasm by performing the complete spasm provocation tests after the administration of nitrates to relieve provoked spasm in the first attempt.