Science and Technology Parks (STPs) are often used as tools to foster regional development. They seek innovations, innovators and encourage innovation amongst the constituent firms, including by networking and knowledge spill over between the inhabitants, Universities and sources of capital. The low success rate of STPs led us to investigate how STP architecture can best cope with a changing and challenging innovation environment, through start-up to early maturity and full maturity, in a preliminary effort to arrive at an evidence-based scheme to help avoid failure. Three different types of architecture were investigated: open (market), star (hierarchy), and closed strong (adhocracy, ambidextrous). Open (market) architecture suffered both from high transaction costs while not protecting against poor decision-making. Results show that it is very beneficial to have a central Cluster Initiative (CI) controlling the decision-making process (star, hierarchy) in the early stages of STP development, where potential gains and losses are relatively modest. However in the early maturity stage with commitment to a high-growth trajectory, a high quality of decision-making is required amongst managers and decisions are best taken by the CI with the input of optimally two individual on-cluster firms. The situation where CI is supported by goodquality decisions from on-cluster firms-an ad hoc, ambidextrous situation-is superior when good innovations abound and the STP has acquired some maturity. However, in environments with a surfeit of poor-fit innovations, this becomes a high-risk strategy with high potential losses and indeed in this situation, retaining a hierarchical (CI only) decision process is most helpful, even when the quality of decision-making amongst CI managers is poor. Results indicate that success involves attracting enough small innovative firms which-in turn-attract larger firms, whose detailed sector-relevant insight improves CI decision-making.