Psychologists have primarily investigated scientific creativity from 2 contrasting in vitro perspectives: correlational studies of the creative person and experimental studies of the creative process. Here the same phenomenon is scrutinized using a 3rd, in vivo perspective, namely, the actual creative products that emerge from individual scientific careers and communities of creative scientists. This behavioral analysis supports the inference that scientific creativity constitutes a form of constrained stochastic behavior. That is, it can be accurately modeled as a quasi-random combinatorial process. Key findings from both correlational and experimental research traditions corroborate this conclusion. The author closes the article by arguing that all 3 perspectives--regarding the product, person, and process--must be integrated into a unified view of scientific creativity.