In this paper we introduce a notion of symplectic normal crossing divisor V and define the GW invariant of a symplectic manifold X relative to such a divisor. Our definition includes normal crossing divisors from algebraic geometry. The invariants we define in this paper are key ingredients in symplectic sum type formulas for GW invariants, and extend those defined in our previous joint work with T.H. Parker [16], which covered the case V was smooth. The main step is the construction of a compact moduli space of relatively stable maps into the pair (X, V ) in the case V is a symplectic normal crossing divisor in X.In previous work with Thomas H. Parker [16] we constructed the relative GromovWitten invariant GW (X, V ) of a closed symplectic manifold X relative to a smooth "divisor" V , that is a (real) codimension 2 symplectic submanifold. These relative invariants are defined by choosing an almost complex structure J on X that is compatible with both V and the symplectic form, and counting J-holomorphic maps that intersect V with specified multiplicities. An important application is the symplectic sum formula ✩ maps, or those that are (C * ) m -equivariant, and all come with corresponding Gromov-type extensions J V. The restriction to level zero induces a fibrationand the rescaling process defines a family of sections (ω ε , J) in it. So J V(X m , V m ) can also be thought as an extension of the parameter space J (X, V ) indexing deformations of Eq. (2.5) for a map f : C → X m , giving rise to a universal moduli space M(X m , V m ) → J V(X m , V m ) consisting of maps into X m satisfying the refined matching conditions (7.6) along the singular locus. As mentioned in Remark 4.19, these parameter spaces come in various flavors, depending on how much of the structure of X m → X they are required to preserve. When restricted to the subspace of parameters which are invariant under the (C * ) m action on the positive levels, we also get a corresponding 'rubber' moduli space M τ (X m , D m )/T τ that appears in (7.9) and which describes the level m stratum of the relative moduli space M(X, V ).Remark A.7. When using Banach norms on the parameter spaces J V (rather than the Frechet topology C ∞ ), the process of rescaling loses m derivatives: e.g. the 'section' aboveHowever, for each s we have an a priori topological bound on the maximum number M of levels entering into the compactification M s (X, V ). So for the purpose of transversality we can start with Sobolev norms W l+M,p on the parameter space J (X, V ) and then take the limit as l → ∞ to get a smooth model for the resolution of each stratum of M s (X, V ).For a family X → B of deformations of a level m building, there is a space J (X → B) of parameters compatible with both the fibration and the total divisor. Since p is a singular fibration, this means that their restriction over each open stratum of B (over which p is a fibration) is compatible with p in the sense of (A.27). The compatibility with the total divisor means that they lift to a pair of parameters ...