We study magnetic field effect on chiral phase transition in a Nambu-Jona-Lasinio model. In comparison with mean field approximation containing quarks only, including mesons as quantum fluctuations in the model leads to a transition from inverse to delayed magnetic catalysis at finite temperature and delays the transition at finite baryon chemical potential. The location of the critical end point depends on the the magnetic field non-monotonously.PACS numbers: 21.65. Qr, 25.27.Nq, 75.30.Kz, 11.30.Rd The research on Quantum Chromodynamics (QCD) phase structure at finite temperature is recently extended to including external magnetic field, due to its close relation to high energy nuclear collisions and cosmological phase transitions. From lattice simulations of QCD at magnetic field eB < 1 GeV 2 ∼ 55 m 2 π [1-4], while the chiral condensate is enhanced at low temperature which is called magnetic catalysis (MC), it is reduced at high temperature which leads to a decreasing critical temperature of chiral phase transition, named as inverse magnetic catalysis (IMC). Many scenarios are proposed to understand the MC and IMC effects. A straightforward question is how the chiral condensate behaves when the magnetic field increases further? Different hypotheses are recently introduced to study the chiral phase transition at extremely strong magnetic field. In some calculations the critical temperature turns around and increases when the magnetic field is sufficiently strong [15,[22][23][24]28], which is named as delayed magnetic catalysis (DMC) [23]. However, it is also argued that the critical temperature will keep decreasing [12,16,29] which is supported by recent lattice simulation [30] where the IMC effect prevails up to eB = 3.25 GeV 2 ∼ 180 m 2 π . The magnetic field effect on QCD phase transitions at finite baryon chemical potential plays an important role in understanding the inner structure of compact stars. Because of the notorious sign problem, there is not yet precise result from lattice simulations. At zero temperature, the critical baryon chemical potential is argued to show a non-monotonous dependence on the magnetic field, and the MC effect is dominant only at sufficiently strong magnetic field, see review [9] and the references therein.In this paper, we investigate the magnetic field effect on chiral phase transition in a Nambu-Jona-Lasinio model (NJL) beyond mean field approximation. We follow the theoretical framework developed in Ref. [27] and focus on the DMC effect at finite temperature and baryon chemical potential. In NJL models, at finite temperature, the MC is described at mean field level [8][9][10][11], and the IMC can be realized by including meson contribution to the quark self-energy [12,27]. Since the DMC is predicted at extremely strong magnetic field, a convincing study in a non-renormalizable model with contact interactions depends strongly on the regularization scheme. We will take a covariant Pauli-Villars regularization which formally allows us to do momentum integrations in the w...