This working paper provides evidence about measuring affective components of subjective well-being in official national surveys. It considers the validity of different affective components, and the modes and contexts through which measurements can be elicited. A rapid systematic review and examples from OECD countries underpin suggested recommendations for consideration in updating existing guidance. These are: 1) measure feelings of loneliness, pain, meaning and joy, which have strong validity, 2) use abbreviated day reconstruction method diaries to minimise recall bias and response burden, and 3) measure affect with questions that have short recall periods, contain language people understand, are placed early on in surveys in a random order and are self-administered. Additionally, we suggest 4) measuring clinical conditions like depression and anxiety with validated diagnostic tests when making a trade-off between clinical and other measures. Future research should 5) provide distributions of – and correlations between – affective, evaluative and mental health items.