The relationship between design and science in architecture continually receives attention from scholars, but to our knowledge, no substantial and systematic literature reviews have been undertaken to generate insights into how the relationship between the two has developed in recent times. As most of the related debate remains theoretical, this paper presents evidence acquired through content analysis of selected 782 peer-reviewed papers published in the last four decades, a period in which design requirements continue to grow in complexity and research efforts increasingly concentrate on design methods. In response, this study departs from the two keywords, architectural science and building science, and develops search strings to support examination of the relationship between science and architectural design since the 1980s, when research that addresses the building environment started to intensify. The analysis employs tools based on machine learning algorithms for text analytics in combination with traditional content analysis methods. The presented research method is fully replicable and stands as one of the contributions of this study. The findings capture the expansion of the research context concerned with the built environment through a careful selection of defining terms, and generate an informed discussion on how the role of science in the architectural design process has changed.