Methodological reductionism has served biology well, but its problems in the study of behaviour include turning open systems into closed ones, defining the units of analysis, and interpreting correlative and causal relationships between processes studied within different biological discourses, from molecular biology to psychology The problems become more acute when methodological becomes philosophical reductionism, with its declared goal of collapsing 'higher level' explanations into 'lower level' ones. Quite apart from the vexed question of what constitutes a 'level', relevant behavioural phenomena may only be manifest at such higher levels. The reductionist programme assumes that parts have ontological and possibly historical (developmental, evolutionary) primacy over wholes, yet the nature of living systems is such that this cannot be the case. I will exemplify these problems in the context of the study of behaviour. But the worst problem arises when reductionism becomes an ideology, especially in the context of human behaviour, when it makes the claims to explain complex social phenomena (e.g. violence, alcoholism, the gender division of labour or sexual orientation) in terms of disordered molecular biology or genes. In doing so, ideological reductionism manifests a cascade of errors in method and logic: reification, arbitrary agglomeration, improper quantification, confusion of statistical artefact with biological reality, spurious localization and misplaced causality.