I review the present status of nonperturbative studies on the nature of the finite temperature QCD transition on the lattice, with the special attention to the determination of the order of the transition. [Talk given at the International Symposium on {\it Origin of Matter and Evolution of Galaxies in the Universe}, Jan.\ 18 -- 20, 1996, Atami, Japan.]

The nature of QCD phase transition is studied with massless up and down quarks and a light strange quark, using the Wilson formalism for quarks on a lattice with the temporal direction extension $N_t=4$. We find that the phase transition is first order in the cases of both about 150 MeV and 400 MeV for the strange quark mass. These results together with those for three degenerate quarks suggest that QCD phase transition in nature is first order.

The deconfining chiral transition in finite-temperature QCD is studied on the lattice using Wilson quarks. After discussing the nature of chiral limit with Wilson quarks, we first study the case of two degenerate quarks $N_F=2$, and find that the transition is smooth in the chiral limit on both $N_t=4$ and 6 lattices. For $N_F=3$, on the other hand, clear two state signals are observed for $m_q \simm{<} 140$ MeV on $\nt=4$ lattices. For a more realistic case of $N_F=2+1$, i.e.\ two degenerate u and d-quarks and a heavier s-quark, we study the cases $m_s \simeq 150$ and 400 MeV with $m_u = m_d \simeq 0$: In contrast to a previous result with staggered quarks, clear two state signals are observed for both cases, suggesting a first order QCD phase transition in the real world.

The finite temperature transition in QCD is studied using Wilson quarks for the cases of $N_F=2$, 3 and 2+1. For $N_F=2$ the transition is smooth in the chiral limit on both $\nt=4$ and 6 lattices. For $N_F=3$, clear two state signals are observed for $\beta\leq4.7$ on $8^2\times10 \times4$ and $12^3\times4$ lattices, which implies the transition is first order for $m_q \simm{<} 140$ MeV. For $N_F=2+1$ we study two cases of $m_s \simeq 150$ and 400 MeV with $m_u=m_d \simeq 0$. In contrast to a previous result with staggered quarks, two state signals are clearly observed for both cases, suggesting a first order QCD phase transition in the real world.

We calculate the tension $\sigma$ of the interface between the confined and deconfined phases by the histogram method in SU(3) lattice gauge theory for temporal extents of 4 and 6 using the recent high-statistics data by QCDPAX collaboration. The results are $\sigma/T_c^3 = 0.0292(22)$ and 0.0218(33) for $N_t=4$ and 6, respectively. The ratio $\sigma/T_c^3$ shows a scaling violation similar to that already observed for the latent heat $\latent$. However, we find that the physically interesting dimensionless combinations $(\sigma^{3}/\latent^2 T)^{1/2}$ and $\sigma T/ \latent$ scale within the statistical errors.