Environmental assessment is an intriguing policy phenomenon: it is employed in an ever-increasing range of contexts the world-over, yet research indicates it rarely efficiently or effectively achieves its principal purpose of promoting sustainable development. Increasingly, practical limitations in the effectiveness of environmental assessment are attributed to its theoretical shortcomings, particularly in relation to the conception of causation. This research advanced debate on environmental assessment by examining the theoretical implications of an empirical analysis of its causal operation in purposefully selected cases. The causal models derived from the research data illustrate the diversity of mechanisms through which environmental assessment can contribute to sustainability, and provide an insight into the complexity and contextuality of causation in the empirical realm. The research findings also highlight a multiplicity of societal expectations concerning environmental assessment's contribution to sustainable development. It is concluded that the interplay of non-rational variables (such as power, agency, experiences and expectations) necessitates the pursuit of a reflexive accommodation of purposes, methods and context in environmental assessment practices.