Purpose: Investigate the behaviour of the cross-sectional stock momentum anomaly in market states. Originality: Methodological setup designed for a more thorough treatment of the topic than existing literature: • Demonstration that market state definition in common use is flawed and results in spurious conclusions, and implementation of an alternative measure; • Examination of role of market state both before and after formation; • Define momentum strategies in a form that allows market states to be uniquely assigned, and which avoid econometric problems described in the literature; • Risk adjustment that respects the inherently time-varying nature of momentum returns. Key literature / theoretical perspective: Motivated by literature seeking to explain momentum anomaly within risk-based or behavioural frameworks. Design/methodology/approach: See “Originality” above. Findings: In contrast to a body of prior literature, momentum returns are not strongly linked to market state; Momentum returns are linked to turning points. Research limitations/implications: Findings reduce support for certain behavioural explanations of momentum anomaly. Practical and Social implications: Implications for practitioner implementation of momentum strategies.