A set $S\subseteq V$ of vertices is an offensive alliance in an undirected graph $G=(V,E)$ if each $v\in N(S)$ has at least as many neighbours in $S$ as it has neighbours (including itself) not in $S$. We study the classical and parameterized complexity of the Offensive Alliance problem, where the aim is to find a minimum size offensive alliance. Our focus here lies on natural parameter as well as parameters that measure the structural properties of the input instance. We enhance our understanding of the problem from the viewpoint of parameterized complexity by showing that (1) the problem is W[1]-hard parameterized by a wide range of fairly restrictive structural parameters such as the feedback vertex set number, treewidth, pathwidth, and treedepth of the input graph; we thereby resolve an open question stated by Bernhard Bliem and Stefan Woltran (2018) concerning the complexity of Offensive Alliance parameterized by treewidth, (2) unless ETH fails, Offensive Alliance problem cannot be solved in time $\mathcal{O}^{*}(2^{o(k \log k)})$ where $k$ is the solution size, (3) Offensive Alliance problem does not admit a polynomial kernel parameterized by solution size and vertex cover of the input graph. On the positive side we prove that (4) Offensive Alliance can be solved in time $\mathcal{O}^{*}(\tt{vc(G)}^{\mathcal{O}(\tt{vc(G)})})$ where $\tt{vc(G)}$ is the vertex cover number of the input graph. In terms of classical complexity, we prove that (5) Offensive Alliance problem cannot be solved in time $2^{o(n)}$ even when restricted to bipartite graphs, unless ETH fails, (6) Offensive Alliance problem cannot be solved in time $2^{o(\sqrt{n})}$ even when restricted to apex graphs, unless ETH fails. We also prove that (7) Offensive Alliance problem is NP-complete even when restricted to bipartite, chordal, split and circle graphs.