A Coupled Map Lattice, which simulates gene expression dynamics inside cells and cellular interactions on a regular lattice, shows a complex pattern of temporal behaviour. The model is represented as a network of genes interacting through their products in space and time in a lattice of genetically identical cells. Despite the fact that the system is described through a step function that imposes a simple repertoire of constant or oscillatory steady states, the dynamics over the lattice are extremely complex. One of the main feature of the asymptotic dynamics is the appearance of long transients in certain regions of parameter space, before the attainment of the final stable attractor. These dynamics, that can grow linearly or exponentially with lattice size, can become the only dynamics computationally observable. The study of the global dynamics-i.e. the average value of the variable over the lattice-shows a qualitative different behaviour depending on the region of the parameter space observed. In the case of the linear transient-growth region the system shows an average that falls quickly on a periodic attractor. In the exponential region values of the average quantities show a behaviour that has stochastic properties. At the boundary of these two regimes the system has an average that shows a complex behaviour before attainment of the final attractor. The possible implications of these results for the study of the dynamical aspects of gene regulation, biochemical pathways and in signal transduction in experimental systems are discussed.